trade performance and competitiveness: selected issues relevant for asian developing economies

8/11/2019 Trade performance and competitiveness: Selected issues relevant for Asian developing economies

http://slidepdf.com/reader/full/trade-performance-and-competitiveness-selected-issues-relevant-for-asian-developing 1/215

ARTNeT

TRADE PERFORMANCE AND COMPETITIVENESS:SELECTED ISSUES RELEVANT FOR

ASIAN DEVELOPING ECONOMIES

A study by the Asia-Pacific Research and Training Network on Trade



The secretariat of the ESCAP is the regional development arm of the United Nations and

serves as the main economic and social development centre of the United Nations in Asia

and the Pacific. Its mandate is to foster cooperation between its 53 members and

9 associate members. It provides the strategic link between global and country-level

programmes and issues. It supports governments of countries in the region in consolidatingregional positions and advocates regional approaches to meeting the region’s unique

socio-economic challenges in a globalizing world. The ESCAP secretariat is located in

Bangkok, Thailand. Please visit the ESCAP website at www.unescap.org for further

information.

Asia-Pacific Research and Training Network on Trade (ARTNeT) is an open regional

network of research and academic institutions specializing in international trade policy and

facilitation issues. Network members currently include over 40 leading national trade

research and academic institutions from 23 developing countries from East, South, and

Southeast Asia and the Pacific. IDRC, UNCTAD, UNDP, ESCAP and the WTO, as core

network partners, provide substantive and/or financial support to the network. The Trade

and Investment Division of ESCAP, the regional branch of the United Nations for Asia andthe Pacific, provides the Secretariat of the network and a direct regional link to trade

policymakers and other international organizations. For more information, please contact

the ARTNeT Secretariat at [email protected] or visit artnet.unescap.org.

The shaded areas of the map are ESCAP members and associate members.

ARTNeT



ARTNeT

Trade performance andcompetitiveness: Selected

issues relevant for Asian

developing economies



TRADE PERFORMANCE AND COMPETITIVENESS:

SELECTED ISSUES RELEVANT FOR ASIAN DEVELOPING

ECONOMIES

A STUDY OF THE ASIA-PACIFIC RESEARCH AND TRAINING NETWORK ON TRADE

United Nations publication

Sales No. E.14.II.F.3

Copyright @ United Nations 2013

All rights reserved

Manufactured in Thailand

ISBN: 978-92-1-120673-9

eISBN: 978-92-1-056600-1

ST/ESCAP/2686

Reference to dollars ($) are to United Stated dollars unless otherwise stated.

Disclaimer

The designations employed and the presentation of the material in this publication do notimply the expression of any opinion whatsoever on the part if the Secretariat of the United

Nations concerning the legal status of any country, territory, city or area of its authorities, or

concerning the delimitation of its frontiers or boundaries.

The opinions, figures and estimates set forth in this publication are the responsibility of the

authors and should not necessarily be considered as reflecting the views or carrying the

endorsement of the United Nations.

Mention of firm names and commercial products does not imply the endorsement of the

United Nations.

All material in this publication may be freely quoted or reprinted, but acknowledgment is

required, together with a copy of the publication containing the quotation or reprint.

The use of this publication for any commercial purpose, including resale, is prohibited

unless permission is first obtained from the secretary of the Publication Board, United

Nations, New York.

Request for permission should state the purpose and the extent of the reproduction.

This publication has been issued without formal editing.



iii

Foreword

This volume comprises four selected research essays on different aspects of trade

and industry issues in the Asian region.1 They are written by researchers in Asian

countries, brought together by the Asia-Pacific Research and Training Network on Trade(ARTNeT) which from April 2011 until September 2013 has implemented its Phase III under

the topic of “Drivers of Competitiveness and Strategies for Economic Diversification for

Developing Countries – Building Policy Making Capacity in Asia and the Pacific”. The latest

phase of ARTNeT reflects the diversity and breadth of research interests, as well as the

depth of research talent, available in the region and harnesses these assets in addressing

the major development policy issues facing developing Asian countries. Since its

establishment ARTNeT has focused the efforts of the network on ensuring that research

institutions of the Asia-Pacific region, particularly those in least developed countries,

provide more relevant and higher-quality applied research and policy recommendations topolicymakers. ARTNeT and its members and associates have endeavored to produce high

quality and relevant studies on trade issues on the basis of a demand-driven research

programme, to improve the communication and dissemination of research study results of

research institutions to policymakers; and to increase the capacity of research institutions in

the region, especially in the least developed countries. All of this has been done with the

aim of making trade and investment related research more useful and accessible to

policymakers who are thus better informed and able to design and prepare for

implementation of coherent trade and investment related policies for inclusive development.

With the success of the previous phases ARTNeT has evolved to be one of the

leading networks of researchers, analysts and policymakers in the region. Its acceptance

as an active and unique network seems fully justified when viewed from its membership

which is cohesive, professionally credible and mutually supportive. The growth and

extensive outputs of the network in itself speaks volumes about the relevance of the

network and the demand for its quality outputs. The interest in the capacity building and

training conducted by ARTNeT has grown at a similar pace and it testifies to the strong

need for organizations like ARTNeT to continue addressing gaps in capacity within the

Asia-Pacific.

The studies in this volume and other research produced under the ARTNeT

Phase III Research programme (available through ARTNeT’s website) are produced with

the aim of addressing these explicit and implicit demands of the Asia-Pacific countries and

to fill in the gaps in capacity and knowledge. The studies are of significance and interest to

researchers and policy analysts alongside policymakers. This is primarily because they

look at previously unexplored issues using new and innovative analytical and

methodological approaches, which can be drawn on to conduct similar studies in the

region. We hope that these will stimulate further work on these important issues.

1 The volume carries only several of the studies that were conducted under this programme phase reflecting

the efforts towards reducing the volume of printed publications and a gradual shift to online release only. The

other studies undertaken under this phase are available in electronic copies.



iv

Acknowledgments

The authors of the studies and the ARTNeT secretariat received invaluable

research guidance, constructive comments and suggestions from advisor to this project

Professor Sisira Jayasuriya, Monash University, Australia, and other experts whom theauthors have acknowledged for specific studies. The project was coordinated by Dr. Mia

Mikic, Chief, Trade Policy and Analysis Section, ESCAP under the general guidance of

Dr. Ravi Ratnayake, Director, Trade and Investment Division, ESCAP. The authors have

also had the opportunity to discuss their methodology and preliminary findings during the

ARTNeT conference on “Empirical and policy issues of integration in Asia and the Pacific”

on 1 to 2 November 2012 held in Colombo, Sri Lanka, in which the discussants, Mr. Denis

Hew, Mr. Jason Lao, Mr. Sarath Rajapatirana, Dr. Nagesh Kumar and other participants

provided very useful comments and suggestions. Notwithstanding this, any remaining

errors and omissions in this publication should be attributed to the authors only and not tothe advisors and benevolent commentators. There are many other people that needed to

be thanked, without implicating them in the views expressed in this volume, for assisting

through the process, since the design of ARTNeT Phase III until preparation of these

selected studies for the publication. These include Ms. Melanie Ramjoue who contributed

to the formulation of the Phase III programme and Mr. Adam Heal who has coordinated the

preparation of the book for print. The studies in the book were edited by Ms. Nicole Colmar

and other research under the programme by Mr. Robert Oliver both of whom should

be thanked for more uniform style of the material presented but should not be blamed

for the content remaining. Ms. Panjai Limchupong, Ms. Tavitra Ruyaphorn andMs. Chaveemon Sukpaibool have all contributed in different but important ways towards

the fruition of the project. Mr. Teemu Alexander Puutio has provided technical support to the

ARTNeT Secretariat and also, with Yisi Chen, the idea for the cover page of the book. The

International Development Research Centre’s support since the establishment of ARTNeT

and during this phase is gratefully acknowledged. This support proved to have been the

factor that made an important difference in the effectiveness of research capacity-building

in the trade and trade-related areas under ARTNeT since 2004. This, combined with

a significant substantive contribution from the ESCAP secretariat as well as a technical

support in the implementation of the projects, have been instrumental in the growingpresence of ARTNeT in the region.



v

Contents

Page

Foreword ................................................................................................................ iii

Acknowledgments ................................................................................................ iv

List of contributors ............................................................................................... xi

Acronyms and Abbreviations .............................................................................. xiii

Overview ................................................................................................................ xv

I. Structural transformation and trade policy: Case of Nepal .................... 1

Paras Kharel

A. Review of literature on structural transformation .................................... 2

B. Nepal’s economic and export performance: A preliminary analysis ....... 9

C. Structural change and productivity growth in Nepal’s economy ............. 19

D. Nepal’s trade and industrial policies in the context of structural

transformation ........................................................................................ 24

E. Methodology for assessing structural transformation through the

“export” lens ........................................................................................... 30

F. Analysis of export baskets and identified products ................................ 34

G. Analysis of targeted products ................................................................. 45

H. Discussion .............................................................................................. 57

Conclusion ..................................................................................................... 66

References .................................................................................................... 69

Appendix ....................................................................................................... 73

II. Logistics performance and trade: An analysis of India’s trade inintermediates with Bangladesh and Thailand .......................................... 91

Prabir De and Amrita Saha

A. Logistics services and production networks: Literature review .............. 92

B. Data and methodology ........................................................................... 95

C. India’s trade with Bangladesh and Thailand........................................... 97

D. Measuring logistics performance ........................................................... 103

E. Does improvement in logistical services lead to higher trade inintermediate goods? ............................................................................... 106

Summary and implications ............................................................................ 112



vi

References .................................................................................................... 114

Appendix ....................................................................................................... 117

III. An analysis of export performance of manufacturing and service

sector enterprises in Sri Lanka .................................................................. 131

JeevikaWeerahewa, Sarath S. Kodithuwakku and Rifana Buhary

A. A model to assess determinants of export performance ........................ 132

B. Data description ..................................................................................... 133

C. Characterization of exporting firms versus non exporting firms ............. 134

D. Results of the estimation of Heckman model ......................................... 139

Conclusions and policy implications .............................................................. 142

References .................................................................................................... 143

Appendix ....................................................................................................... 145

IV. Evaluation of business association membership on small and

medium enterprises’ growth performance: Evidence from enterprisesurvey of Cambodia .................................................................................... 147

Vathana Roth

A. Literature review and theoretical aspects ............................................... 149

B. Econometric specifications and variables measurement ....................... 152

C. Data source and descriptive statistics .................................................... 156

D. Regression results and discussion ......................................................... 158

E. Limitations .............................................................................................. 162

Conclusion and areas for future research ..................................................... 162

Annex ............................................................................................................ 164

References .................................................................................................... 188

Contents (continued)

Page



vii

List of Tables

Chapter I

Table 1. Nepal in comparative perspective: Trade performance .......................... 13

Table 2. Decomposition of Nepal’s labour productivity growth during 1999-2008 20

Table 3. Sectoral employment, value added shares and their change, and

productivity level and growth (Nepal) ..................................................... 22

Table 4. Products identified in Trade Policy 2009 ................................................ 25

Table 5. Priority export potential products/sectors identified by Nepal TradeIntegration Strategy 2010 ....................................................................... 26

Table 6. Features of export basket in 2010.......................................................... 40

Table 7. Top 10 nearby products, ordered by Strategic Value ............................. 41

Table 8. Top 10 middle-distance products, ordered by Strategic Value ............... 42

Table 9. Top 10 faraway products, ordered by Strategic Value............................ 43

Table 10. Products in Open Forest according to Leamer classification ................. 44

Table 11. Summary features of NTIS-identified products ...................................... 46

Table 12. Summary statistics of PRODY, Path and Strategic Value of across

groups .................................................................................................... 48

Table 13. NTIS products that lie in the Open Forest .............................................. 49

Table 14. Comparison of NTIS products with export basket and Open Forest ...... 50

Table 15. Summary statistics of Trade Policy identified products .......................... 51

Table 16. Summary features of Trade Policy identified products ........................... 51

Table 17. Distribution of Trade Policy identified products across path terciles ...... 53

Table 18. Number of products exceeding median values of comparator groups ... 54

Table 19. Trade Policy products that lie in the Open Forest .................................. 55

Table 20. Features of “other” NTIS products ......................................................... 56

Table 21. Processed vs. non-processed agricultural products .............................. 63

Chapter II

Table 1. India’s imports of ACE from Thailand and export of yarn toBangladesh: 2000-2010 ......................................................................... 98

Table 2. Intra-Industry trade index (2007): Common set of products at 6-digit HS 100


Page



viii

Table 3. IIT in textile and clothing sector, 2010 (Exporter – India, Importer –

Bangladesh) ........................................................................................... 100

Table 4. IIT in air conditioning equipment (Importer – India, Exporter – Thailand) 101

Table 5. Vertical trade potential between India and Bangladesh ......................... 102

Table 6. Vertical trade potential between India and Thailand .............................. 103

Table 7. LPI scores and ranks ............................................................................. 104

Table 8. Baseline regression (OLS): Fixed effect model ...................................... 106

Table 9. Arellano-Bover dynamic panel-data estimation (System GMM) ............ 108

Table 10. Im, Pesaran and Shin (IPS) panel unit root test (Period: 2000-2010) .... 109

Table 11(a). India’s exports of yarn to Bangladesh: Westerlund panel

co-integration test (Period: 2000-2010) ............................................. 110

Table 11(b). India’s exports of yarn to Bangladesh: Westerlund panel

co-integration test (Period: 2000-2010) ............................................. 111

Table 12. Panel Granger causality test between trade and LPI ............................. 112

Chapter III

Table 1. Heterogeneity of firms by size ................................................................ 136

Table 2. Heterogeneity of firms by industry .......................................................... 136

Table 3. Export propensity, intensity by industry and province ............................ 137

Table 4. Descriptive statistics of variables used in the Heckman estimation ....... 139

Table 5. Results of the Heckman Model estimation ............................................. 141

Chapter IV

Table 1. Classification of enterprises in Cambodia .............................................. 164

Table 2. Distribution of participating and non-participating groups by firm size

and sub-sector ....................................................................................... 164

Table 3. Number of establishments as exporter .................................................. 165

Table 4. Membership in business association or chamber of commerce by

firm’s characteristics ............................................................................... 166

Table 5. Firm’s characteristics and performance ................................................. 168

Table 6. Perceived important services members expect from BA or CoC ........... 171

Table 7. Firm’s years of experience since establishment and performance ........ 171


Page



ix

Table 8. Descriptive statistics of main outcome and indicator variables for all

firms ....................................................................................................... 172

Table 9. Descriptive statistics of main outcome and indicator variables for SMEs 174

Table 10. Logit regression to estimate propensity score of membership

participation conditional on observed and selected coveriates .............. 176

Table 11. PSM estimation results of memberships in business associations or

chambers of commerce (pooled) ........................................................... 181

Table 12. PSM estimation results of memberships in business associations or

chambers of commerce (SMEs) ............................................................. 182Table 13. OLS estimation results of memberships in business associations or

chambers of commerce .......................................................................... 183

Table 14. PSM & OLS estimation results of memberships in business

associations or chambers of commerce................................................. 184

List of Figures

Chapter I

Figure 1. The product space ............................................................................... 4

Figure 2. Nepal per capita GDP growth performance ......................................... 10

Figure 3. Nepal in comparative perspective: Growth of GDP per capita............. 10

Figure 4. Nepal in comparative perspective: Manufacturing, value added

(per cent of GDP), 2009 ....................................................................... 11

Figure 5. Share of manufacturing in Nepal’s GDP over time .............................. 12

Figure 6. Nepal’s trade performance................................................................... 12

Figure 7. Nepal’s exports of goods and services ................................................ 13

Figure 8. Nepal’s export composition (in terms SITC 1-digit categories) ............ 14

Figure 9. Nepal’s export composition in terms of Leamer categories ................. 15

Figure 10. Comparative export composition – Leamer categories 2010 .............. 17

Figure 11. Major exports ....................................................................................... 17

Figure 12. Nepal’s export composition – Lall classification ................................... 18

Figure 13. Nepal in comparative perspective: Export composition – Lall

classification, 2010 .............................................................................. 19


Page



x

Figure 14. Correlation between sectoral productivity and change in employment

shares in Nepal (1999-2008) ............................................................... 21

Figure 15. Evolution of EXPY ............................................................................... 35

Figure 16. Nepal in comparative perspective: Export sophistication 2010 ........... 36

Figure 17. Nepal’s export sophistication ............................................................... 37

Figure 18. Nepal in comparative perspective: Export sophistication 2010 ........... 37

Figure 19. Strategic Value against density of unexploited products, 2010............ 44

Figure 20. Sophistication and connectivity of NTIS-identified products ................ 47

Figure 21. Sophistication and connectivity of Trade Policy products .................... 53

Chapter II

Figure 1. Production blocks and logistics service links ....................................... 94

Figure 2. Trends in trade shares: India’s export of yarn to Bangladesh and

import of ACE from Thailand ................................................................ 98

Chapter III

Figure 1(a). Exporters vs non-exporters by size ..................................................... 135

Figure 1(b). Exporters vs non-exporters by province .............................................. 135

Figure 1(c). Exporters vs non-exporters by industry ............................................... 135

Figure 2(a).Kernal density of log sale ..................................................................... 135

Figure 2(b). Kernal density of log labour productivity .............................................. 135

Figure 2(c). Kernal density of log wage .................................................................. 135

Figure 3. Export concentration by value of exports and total sales .................... 138


Page



xi

List of contributors

Rifana Buhary is staff at the Department of Agricultural Economics and Business

Management, Faculty of Agriculture, University of Peradeniya, Sri Lanka.

Prabir De is a Senior Fellow at the Research and Information System for Developing

Countries (RIS). De works in the field of international economics and has research interests

in international trade and development. He was a visiting research scholar of the Asian

Development Bank Institute (ADBI), Tokyo; Korea Institute of International Economic Policy

(KIEP), Seoul; and United Nations Economic and Social Commission for Asia and the

Pacific (ESCAP), Bangkok. He is also a visiting faculty member of the Institute of Foreign

Policy Studies (IFPS), Calcutta University, Kolkata, where he has been teaching trade and

regional cooperation. He has been conducting policy research for the Government of India

and several international organizations. De has a Ph.D. in Economics from the Jadavpur University, Calcutta. He has contributed several research papers in international journals

and written books on trade and development. He is the managing editor of South Asia

Economic Journal, published by Sage, and edits a policy brief series, called Mekong –

Ganga Policy Brief. e-mail: [email protected]

Sisira Jayasuriya is a Professor of Economics in the Department of Economics, Monash

University, Melbourne, Australia and an Adjunct Professor at the Arndt-Corden Department

of Economics at the Australian National University, Canberra, Australia. He has held

previous appointments at the Australian National University, La Trobe University, University

of Melbourne and the International Rice Research Institute, Philippines. Hisresearch

interests are in policy-oriented development economics, and he has published extensively

on trade, macroeconomics, resource and environmental issues of developing countries,

with a primary focus on Asia. He has been a consultant to World Bank, Asian Development

Bank, International Food Policy Research Institute, Food and Agriculture Organization

(FAO) and several other UN agencies), and to government agencies in both Australia and

overseas. e-mail: [email protected]

Paras Kharel is a trade economist. His research interests include trade costs, trade

facilitation including transit issues, competitiveness, aid for trade, trade and climate change

linkages, trade and transit issues related to least developed countries (LDCs) and

landlocked developing countries (LLDCs) and services trade. He has worked as

programme officer and senior programme officer at South Asia Watch on Trade, Economics

& Environment (SAWTEE) since 2007. e-mail: [email protected]

Sarath S. Kodithuwakku has obtained his BS in agriculture with specialization in

agricultural economics from the University of Peradeniya, Sri Lanka in 1990. He obtained

his Masters of Business Administration in Marketing and Ph.D. in entrepreneurship from the

University of Stirling, United Kingdom in 1993 and 1997, respectively. He currently servesas a Professor in the University of Peradeniya. He founded the MBA programme offered by

the University of Peradeniya and he has been the Coordinator of the programme since its

inception in 1998, in addition to serving as the Chairman of the Board of Study in Business



xii

Administration, PGIA since its inception in 2002. He was also the Director of the

Agribusiness Centre, the outreach arm of the Faculty of Agriculture, since its inception until

September 2011. At national level, he has been serving as the Senior Vice President of the

Sri Lanka Institute of Management, the premier professional body for Sri Lankan

management professionals since 2003 and as a member of Board of Study of the Sri LankaInstitute of Marketing, the premier professional body for Sri Lankan marketing professionals

since 2002. e-mail: [email protected]

Amrita Saha is a Ph.D. student working on issues in international trade and trade policy.

Her research is funded by the Commonwealth Scholarship Commission, United Kingdom.

Prior to joining the research programme at Sussex, she has worked with the Foreign Trade

Division at Ministry of Commerce, India. She has also pursued several research projects in

India, under the MacArthur fellowship at the Indian Council for Research in International

Economic Relations (ICRIER) New Delhi, the Sir Ratan Tata Fellowship at the Institute for

Social and Economic Change (ISEC) Bangalore, at the Reserve Bank of India (RBI)

Bangalore and the National Council for Applied Economic Research (NCAER) New Delhi.

She has completed Bachelors (Honours) in Economics from Delhi University, India and

Master Degree in Economics from the University of Hyderabad, India.

Jeevika Weerahewa is a Professor at the Department of Agricultural Economics and

Business Management, Faculty of Agriculture, University of Peradeniya, Sri Lanka. She

has a B.Sc. and M.Phil. in Agriculture from the University of Peradeniya and a Ph.D. in

Agricultural Economics from the University of Guleph, Canada. She has been teaching

fundamental and applied courses in Economics with a special focus placed on agricultureand allied fields at the University of Peradeniya since 1987. Her primary research area is

Quantitative Policy Analysis and applications are extended to analysis of various economic

and distributional impacts of trade, agricultural production and marketing policies. She is

a Collaborator of the International Food Policy Research Institute (IFPRI), a Hewlett Fellow

of the International Agricultural Trade Research Consortium (IATRC), a Fellow of the

Canadian Agricultural Trade Policy Research Network (CATPRN) and a Focal Point of the

Asia-Pacific Research and Training Network on Trade (ARTNeT). e-mail: jeevikaw@

gmail.com

Vathana Roth is a research associate at Cambodia Development Resource Institute

working on issues such as economic growth, poverty, inequality and private sector

development. He is also a lecturer of economics at the Department of International Studies,

Institute of Foreign Languages. He previously served as Assistant to Deputy Country

Director (Programme) of the United Nations Development Programme, Cambodia. e-mail:

[email protected]



xiii

Acronyms and Abbreviations

AB Arellano-Bond

ACE Air-Conditioning Equipment

ADB Asian Development Bank ADF Augmented Dickey-fuller

ARTNeT Asia-Pacific Research and Training Network on Trade

ASEAN Association of Southeast Asian Nations

ATE Average Treatment Effects

BAs Business Associations

BEC Broad Economic Categories

CAGR Compound Annual Growth Rate

CES Constant Elasticity of Substitution

CoC Chamber of CommerceDCS Department of Census and Statistics of Sri Lanka

DID difference-in-differences

EHS Early Head Start

ERIA Economic Research Institute for ASEAN and East Asia

ESCAP Economic and Social Commission for Asia and the Pacific

EXPY export basket of a country

FEM Finite Element Method

FTAs Free Trade Agreements

GDP Gross domestic productHS Harmonized System

CIC Industry and Commerce

ICT Information and Communication Technology

IDE-JETRO Institute of Developing Economies-Japan External Trade Organization

IIT Intra-Industry Trade

IoO Index of Opportunities

ISIC International Standard Industrial Classification of All Economic Activities

IT Information technology

ITC International Trade CentreLDCs Least Developed Countries

LPI Logistics Performance Index

MNEs Multinational Entities

MSMEs Micro, small, and medium Enterprises

NAFTA North America Free Trade Agreement

NBER National Bureau of Economic Research

NIC National Industrial Classification

NTIS Nepal Trade Integration Strategy

OLS Ordinary Least SquaresPCA Principal Component Analysis

Pce per capita consumption of electricity

PSM Propensity score matching



xiv

RCA Revealed comparative advantage

SAFTA South Asian Free Trade Area

SEZ Special Economic Zone

SITC Standard International Trade Classification

SMEs Small and Medium EnterprisesSSCs Supply-side constraints

SV Strategic Value

UNCTAD United Nations Conference on Trade and Development

USDA United States Department of Agriculture

WDI World Development Indicators

WITS World Integrated Trade Solution

WTO World Trade Organization



xv

Overview

The first essay on “Structural transformation and trade policy: The Case of Nepal”

by Paras Kharel reviews recent trade and industry policy initiatives in Nepal. This is in the

context of recent literature pioneered by Dani Rodrik, Ricardo Hausmann and colleagues,suggesting that some patterns of production and trade may be more conducive to structural

transformation and industrialization. The essay provides an excellent overview of this new

literature, which is starting to have an important influence in both academic and policy

circles. It also provides an application of some of the approaches using network analysis to

develop a graphical representation of the product space in Nepal. Based on a very detailed

and comprehensive analysis, Kharel draws important implications for trade and industry

policy. He argues that there is scope for better targeting of Nepalese trade and industry

promotion policies.

Kharel’s discussion raises some of the unresolved and difficult issues in targeted

government interventions of this type. These issues have a long history; from the time

when import substitution industrialization (ISI) policies held sway in much of the developing

world, and that continue today to have relevance for policy debates. This is not only in

Nepal, but also for many other countries grappling with the challenges of sustaining

industrialization and development over the longer term. The long history of ISI policies

provides plenty of examples where government interventions to guide the industrialization

process through targeted support has resulted in costly misallocation of resources, which

often trap economies in “infant industries that never grow up”. While making a case for governments to pay attention to the potential benefits of well-directed industry promotion,

Kharel also points to the potential tension between assisting industries with the highest

potential for export earnings and employment and, on the other hand, assistance for

upgrading the industrial and export structures, and capabilities to be able to produce and

export (or expand) the production and export of, more sophisticated products requiring

a greater number of capabilities. For a country like Nepal, traditional labour-intensive

industries (such as garments) may offer the best potential in terms of maximising export

earnings and employment. However, it may not be most helpful for industry upgrading and

a progressive shift into more sophisticated manufactured products.

In the second essay, “Logistics performance, trade and production fragmentation:

An analysis of India’s trade with Bangladesh and Thailand”, Prabir De and Amrita Saha

investigate the important issue of logistical services for successful industrialization and

export growth. They analyse the link between logistical performance and trade in parts and

components in the manufacturing industries. This is central to the growth of international

production linkages and networks associated with the phenomenon of international

production fragmentation. They focus specifically on trade in two commodities – India’s

export of yarn to Bangladesh and India’s import of air-conditioning equipment from

Thailand, in the context of the broader regional trading context.

Logistics involves a multiplicity of services including various forms of transport,

communications and quality of human resources. They develop a logistical services index



xvi

for 20 Asia-Pacific countries based on data on a wide range of such services. This is to

quantify the link between logistic services and trade. They demonstrate not only that

improved logistical services have a positive impact on trade, but also that there is a two-

way relationship between trade and logistical services in that increased trade, in turn, tends

to enhance the quality of logistical services. Although the precise mechanism throughwhich this occurs is not explored in the paper, it is likely that expanded trade in turn may

induce market and government policy responses. The wide differences observed in quality

of logistical service among the countries studied indicate that there is a large potential for

improvement in logistics. In turn, this contributes to beneficial expansion of trade and

fosters closer and deeper regional production linkages.

The pace and impact of progress in trade and industrialization in the region

ultimately depends on the extent to which firms can address the challenges of competing

and succeeding in markets where policy liberalization and the resulting pressures from

globalization intensify competitive pressures, even as they open up broader opportunities.

The importance of firm level differences in critically influencing outcomes when economies

are opened up to international competition has been highlighted in recent literature on

international trade following Melitz (2003). The next two essays explore firm level issues

related to competing in globalized markets.

In the third essay, “An analysis of export performance of manufacturing and service

sector enterprises in Sri Lanka”, Jeevika Weerahewa, Sarath S. Kodithuwakku and Rifana

Buhary investigate the different characteristics of exporting and non-exporting firms in two

key industries using a rice data set from a recent (2011) World Bank “Enterprise Survey” of over 800 firms. The researchers chose econometric techniques and a modelling approach

that avoids the common problem of “selection bias” in studies of this type. They show that

several firm-level characteristics, in particular size and geographical location, as well as the

managers’ perceptions regarding the overall business environment, influence whether firms

become exporters or not. Indeed, a few large exporters, many with foreign linkages,

account for the bulk of exports. The results of this study raise several issues that merit

future investigation in more detail, including the reasons for different managerial

perceptions regarding the business climate. The overall impression conveyed by the results

of this analysis is that Sri Lanka faces formidable challenges if it were to succeed infostering a broad based export culture among the small and medium sized firms who

comprise the majority.

The final essay, “Evaluation of business association membership on small and

medium enterprises’ growth performance: Evidence from enterprise survey of Cambodia” is

by Vathana Roth. It addresses issues relating to the major policy challenge of how to assist

small and medium enterprises operating in a developing country setting of imperfect factor

and information markets, and how to overcome scale and market access constraints. It

examines factors that impact on performance among small and medium firms, with a focuson whether business associations can contribute to increasing their competitiveness and

overall performance. The study uses data from the 2007 World Bank Enterprise Surveys of

Cambodia to examine the impact of belonging to a business association on Small and



xvii

Medium Enterprises (SMEs) from four sectors – manufacturing, tourism, trade, and others.

The analytical approach involved propensity score matching (PSM) as well as PSM with

ordinary least squares (OLS) regression. The study finds that firms belonging to business

associations tend to have higher turnover and production, and tend to spend more on

production and other related costs. However, it did not find membership to have anysignificant impact on firms’ labour productivity and labour cost per worker. This may be due

to business associations in Cambodia having limited capacity to enhance productivity of

members through provision of appropriate services. In any case, the results ought to be

treated with some caution, not only because of data limitations but also as participation in

such associations is a relatively new experience for many firms.



1

I. Structural transformation and trade policy:Case of Nepal

By Paras Kharel

Introduction

High and sustained economic growth entails structural transformation. This involves

the shift of productive factors from low-productivity and low-wage activities to

high-productivity and high-wage activities. The aim is to move the output structure into

higher-productivity activities, and production of more complex and sophisticated products.

In the past, structural transformation was generally seen as a consequence of growth anddevelopment. However, recent advance in literature has produced powerful analysis

methods for appreciating the role of structural transformation in inducing growth and

development. It also provides evidence that product potential varies in effecting structural

transformation. In particular, what a country exports now influences the type of goods it will

export in future, thus influencing its future economic growth rate. In this context, the

literature suggests that governments may have a more direct and important role to play,

particularly in economies with low economic complexity.

These findings are particularly relevant for Nepal. Asa least developed economystarting to rebuild its economy after a period of political conflict, it is now attempting to

embark on a sustainable economic development path. The country has formulated

a range of policies, including Trade Policy 2009 and Nepal Trade Integration Strategy

(NTIS) 2010, for active government intervention in stimulating export growth and

industrialization. It has also selected products to be given high-priority status for

government support. This paper analyses the nature and extent of structural change in the

Nepali economy by studying trends, patterns and composition of productivity growth. It also

assesses Nepal’s export performance, including the sophistication and diversification of its

exports, and the extent government high-priority products have in assisting Nepal’sstructural transformation and future growth.

The rest of the paper is organized as follows. Section 2 reviews the literature on

structural transformation, including recent methodological advances. Section 3 discusses

Nepal’s economic and export performance. Section 4 discusses sectoral productivity

growth and whether structural change has been growth-enhancing or growth-reducing in

Nepal over a recent decadal period. Section 5 reviews Nepal’s trade policy, plans and

strategies, with focus on products identified for export promotion. Section 6 discusses the

methods and data used for analysing the nature, pattern and prospects of structural

transformation from the export dimension. Section 7 analyses the evolution of Nepal’s

export basket in terms of export sophistication and diversification. It also assesses the

prospects for structural transformation offered by products identified/targeted by Trade

Policy 2009 and NTIS 2010. Section 8 concludes.



2

A. Review of literature on structural transformation

High and sustained economic growth entails structural transformation – a shift of

productive factors from low-productivity and low-wage activities to high-productivity

and high-wage activities. This is such that the output structure progressively shifts intohigher-productivity activities, including increased production of more complex and

sophisticated products. Countries should be able to produce not just more of the same

products, but also new ones. Also, the products chosen for specialization will have different

consequences for development.

Founders of development economics emphasize the importance of industrialization

for the externalities it generates, leading to accelerated growth (Rosenstein-Rodan, 1943;

Hirschman, 1958; Kaldor, 1967). Experiences of industrialized economies, as well as those

of the newly industrialized East Asia, suggests the importance of structural transformation

(see, for example, Chang, 2002). However, lacking formal models, mainstream economic

theory has made little use of these ideas (Hidalgo and Hausmann, 2008:5).

Dominant neo-classical trade and growth theories state that the type of products

a country produces and exports have little or no bearing on long-term growth and

development. The Heckscher-Ohlin model suggests that, in an open economy, countries

specialize on the production of goods that intensively use the productive factors that they

are endowed with, such as physical capital, labour, land, human capital, infrastructure and

institutions. The specialization pattern changes with the accumulation speed of specific

factors. Therefore, controlling for initial factor endowments, the particular productsa country produces and exports do not matter for its future economic performance

(including export).

Similarly, the Ricardian model argues that technological differences across

countries determine comparative advantage. Also two other dominant theories – the

varieties model of Romer (1990), and the quality ladders model of Aghion and Howitt

(1992) and Grossman and Helpman (1991) – explain productivity differences as “assume

a degree of homogeneity across products that eliminates the possibility to capture the

impact of initial specialization” (Hausmann and Klinger, 2007:1). New trade theory(Helpman and Krugman, 1985; Krugman, 1979) explains intra-industry specialization

(which Ricardian and Hecksher-Ohlin models cannot) through economies of scale and

product differentiation. The so-called “new-new” trade theory – the Melitz model (Melitz,

2003) – takes into account heterogeneity among firms, explaining which firms would find it

advantageous to export and which firms would sell only in the domestic market. However,

neither explains the path-dependent process of specialization. Thus, dominant mainstream

economic theories does not consider the structure of the product space (the universe of

goods and services that a country may produce) to be important for future growth, ignoring

the path-dependent nature of growth and development.

Recent advances in the literature challenge this view with more explicit recognition

of externalities and path dependence. Hausmann et al. (2006) finds that countries that

export goods associated with higher productivity levels grew more rapidly, even after



3

controlling for initial income per head, human capital levels, and time-invariant country

characteristics. This is through measures of the productivity or income potential of

a product (PRODY and EXPY). Their findings also reveal that countries with initial high

levels of export sophistication subsequently experienced higher growth in exports.

Hausmann et al. (2006) argues that “countries become what they produce”. This appeals tothe mechanism of “cost discovery” (Hausmann and Rodrik, 2003), under which the range of

goods that an economy ends up producing and exporting is determined not just by the

usual fundamentals, but also by the number of entrepreneurs who can be stimulated to

discover the cost of production in modern sectors of the economy. Such cost discovery

generates considerable positive externalities for other entrepreneurs.

Hausmann and Klinger (2006, 2007) show that changes in the revealed

comparative advantage of nations are governed by the pattern of relatedness of products at

the global level. As countries change their export mix, there is a strong tendency to move

towards “related goods” (goods that are, somewhat, more sophisticated but fairly similar),

rather than to goods that are farther away. They introduced an outcome-based measure of

relatedness, called proximity, between pairs of products using cross-country export data.

Formally, the proximity between products i and j is the minimum of the pair-wise conditional

probabilities of a country exporting a good, given that it exports another. Their findings

show that the pattern of relatedness of products is only very partially explained by similarity

in broad factor or technological intensity classifications, as in Leamer (1984) or Lall (2000).

This suggests that the relevant determinants are much more product-specific. Countries

that specialize in a dense part of the product space (where there are a lot of products in

close proximity to one another) find it easier to change their revealed comparative

advantage than countries that specialize in more disconnected products.

In general, rich (poor) countries tend to specialize in dense (sparse) parts of the

product space, although there is significant variation in this relationship. Controlling for the

level of income, countries like China, India, Indonesia, Turkey and Poland, specialize in

a very dense part of the product space, while countries that specialize in natural resources

(particularly oil) have export baskets in disconnected parts of the space (Hausmann and

Klinger, 2007:16). Hausmann and Klinger (2007) find that the speed at which countries can

transform their productive structure and upgrade their exports depends on having a path of nearby goods that are of increasingly higher value. In their model, they argue that the

assets and capabilities2 needed to produce one good are imperfect substitutes for those

needed to produce another good. However, this degree of asset specificity varies,

determining product interrelatedness. As a result, the process of structural transformation

tends to favour nearby goods in the product space, making the pattern of structural

transformation path-dependent.

2 The model focuses on human capital but is applicable to other specific non-tradable assets like

knowledge, labour training requirements, infrastructure needs, property rights, regulatory requirements or other

public goods (Hausmann and Klinger (2006, 2007).



4

Building on Hausmann and Klinger (2006, 2007), Hidalgo et al. (2007) uses tools of

network analysis to develop a graphical representation of the product space (the network of

relatedness of products).3 The product space (covering 775 four-digit products of SITC

Rev. 2) is highly heterogeneous, with a core-periphery structure (figure 1).

Figure 1. The product space

Source: Hidalgo et al. (2007)

3 Product space is the network of relatedness between products traded in the world, which is very useful to

study the structural transformation of economies. The product space analysis was pioneered by Hausmannand Klinger (2006). Hidalgo et al. (2007)) uses the tools of network analysis to construct a visual image of the

product space. In product space analysis, the proximity (øij) between products does not measure physical

distance but is a conditional probability-based measure used as an inverse proxy for distance between two

products. It is calculated as the minimum of probability that a country exporting product B with comparative

advantage also exports A with comparative advantage, and vice versa.

Hausmann and Klinger (2006, 2007) and Hidalgo et al. (2007) find that

more-sophisticated products are located in a densely connected core (metal products,

machinery, and chemicals, as per Leamer classification), whereas less sophisticated

products occupy a less-connected periphery. At the top of the periphery are fishing, animal,

tropical and cereal agriculture products. To the left are two strong peripheral clusters

formed by garments and textiles, followed by a second animal agriculture cluster. At the

bottom of the product space is a large electronics cluster and to the right there is mining,

followed by forest and paper products. Poorer countries tend to be located in the periphery.



5

The empirical findings indicate that countries move through the product space by

developing goods close to those they currently produce, and that many countries will meet

considerable difficulty in reaching the core. An important implication of these findings is that

even with similar levels of production and export sophistication, countries face different

prospects for structural transformation. This would depend on the proximity of moresophisticated products to their respective current productive capacities (proxied by the

products that a country exports with revealed comparative advantage).

Felipe et al. (2010a) develops an “Index of Opportunities” (IoO) for 130 countries

based on their capabilities to undergo structural transformation. The four dimensions of IoO

are related to the characteristics of a country’s export basket: sophistication, diversification,

standardness, and possibilities to export other products with comparative advantage. The

rationale behind IoO is that in the long run a country’s income is determined by the

sophistication and variety of products it makes and exports, reflecting its capabilities. In the

IoO rankings, Nepal ranks 33rd among 96 non-high income countries, putting it in the

second quintile. Among the four South Asian countries (for which data is available), Nepal’s

capabilities to undergo structural transformation are better than Bangladesh, Pakistan and

Sri Lanka and second only to India.

Felipe et al. (2010b) argues that becoming a rich country requires the ability to

produce and export goods that embody certain characteristics. In classifying 779 exported

goods (SITC Rev. 2, 4-digit), two dimensions are accorded – sophistication (measured by

PRODY) and connectivity to other products (measured by PATH). As a result, 352 “good”

products and 427 “bad” products are identified.

Researchers then categorized 154 countries into four groups according to these two

characteristics, with Nepal being among the “low core” 75 countries in the “low product”

trap. This group also includes Pakistan, Sri Lanka and Bangladesh from South Asia. The

share of core products in the number of products Nepal exports with comparative

advantage is 18.8 per cent. The exports of the countries in this group are concentrated in

products with low sophistication and little or average linkages with other products. Felipe

et al. (2010b) suggests that to escape this situation, these countries need to implement

policies that would help them accumulate the capabilities needed to manufacture andexport more sophisticated and better connected products. This would involve human capital

acquiring skills, technology and knowledge; a higher drive to diversify and to increase

sophistication by embracing a realistic industrial vision; and improvement in organizational

abilities e.g. firm-level organization (Felipe et al., 2010b: p. 30).

Felipe et al. (2010c) argue that the key factor underlying China’s fast development

during the last 50 years is its “ability to master and accumulate new and more complex

capabilities”, citing the increase in diversification and sophistication of its export basket.

Furthermore, they say China’s accumulation of new capabilities is policy induced and not

the result of the market, beginning before economic liberalization started. Analysis of

China’s current export opportunity set shows that the country is “exceptionally well

positioned” to continue learning and gaining revealed comparative advantage in the export

of more sophisticated products.



6

Abdon and Felipe (2011) find that in contrast the majority of Sub-Saharan African

countries are trapped in the export of unsophisticated, highly-standard products that are

poorly connected in the product space. This makes the region’s process of structural

transformation particularly difficult. As the products that are “nearby” to those they already

export have the same characteristics, Abdon and Felipe (2011) conclude that shifting tothese products will do little to improve the region’s growth prospects. As a means for

jump-starting and sustaining growth, they recommend implementation of policies and

provision of public inputs that encourage the private sector to invest in new and more

sophisticated activities.

Hidalgo and Hausmann (2009) use the techniques of network science to develop

a method to characterize the structure of bipartite networks connecting countries to the

products they export. Labeled Method of Reflections, it creates measures to count the

relative number of capabilities present in a country without making any assumptions about

the nature of capabilities. This is done by iteratively combining information on diversity of

countries (number of products a country exports) and ubiquity of products (number of

countries that export a product) using trade data. The number of capabilities present in

a country forms the country’s economic complexity. The complexity measures developed

through the Method of Reflections do not include information on income.4 This method of

measuring complexity of product and economy addresses criticism of PRODY and EXPY

(measures using income information), which makes the tautological observation that rich

countries export rich country goods.

Hidalgo and Hausmann (2009) find that a) the complexity of a country’s economy iscorrelated with per capita income. Deviations from this relationship are predictive of future

growth, suggesting that countries tend to approach the level of income associated with the

available capability set. B) The level of complexity of a country’s economy predicts the type

of products that a country will be able to develop in the future. This suggests that the new

products that a country develops depend substantially on the capabilities already available

in that country. Hidalgo and Hausmann (2009) argue that changes in a country’s

productive structure can be understood as a combination of two processes a) that by which

“countries find new products as yet unexplored combinations of the capabilities they

already have”, and b) that by which “countries accumulate new capabilities and combinethem with other previously available capabilities to develop yet more products” (Hidalgo

and Hausmann, 2009:10575).

Ranking of products and countries according to the measures of complexity

(developed by Hidalgo and Hausmann (2009); Abdon et al. (2010)) finds that the most

complex products are in machinery, chemicals and metals, while the least complex

products are raw materials and commodities, wood, textiles, and agricultural products.

More so, the most complex economies in the world are Japan, Germany, and Sweden, and

the least complex, Cambodia, Papua New Guinea, and Nigeria; the major exporters of the

4 However, there is a strong correspondence between PRODY and EXPY with their network counterparts,

“suggesting that most of the information contained in PRODY and EXPY comes from the structure of the

network connecting countries to the products they export, rather than from income” [Hidalgo 2009: p. 7].



7

more complex products are the high-income countries, while the major exporters of the less

complex products are the low-income countries; and export shares of the more complex

products increase with income, while export shares of the less complex products decrease

with income. Among the South Asian economies with available data, Nepal is ranked 89 th

out of 124 countries in terms of economic complexity. This is higher than Bangladesh, SriLanka and Pakistan but below India. Despite this, Nepal’s export basket is positively

skewed towards less complex products.

Using the measures of complexity (diversification and ubiquity) developed in 2009,

Hausmann and Hidalgo, in 2010, construct a simple model that assumes that each product

requires a potentially large number of non-tradable inputs (capabilities). More so, that

a country can only make the products for which it has all the requisite capabilities. Within

the model, products differ in the number and specific nature of the capabilities they require,

while countries differ in the number and nature of capabilities they have. Therefore,

products that require more capabilities will be less ubiquitous, while countries that have

more capabilities will be more diversified. Mathematically, this proves that: i) the level of

diversification of a country increases on average with the number of capabilities it has;

ii) the ubiquity of a product decreases, on average, with the number of capabilities it

requires; iii) the average ubiquity of products exported by a country decreases with that

country’s level of diversification; iv) the average level of diversification of products exported

decreases with the ubiquity of that product.

The model also implies that the return (in terms of diversification) to the

accumulation of new capabilities increases exponentially with the number of capabilitiesalready available in a country. This gives rise to the “quiescence trap” or a “trap of

economic stasis”, meaning countries with few capabilities will have negligible or no return to

the accumulation of more capabilities, while countries with many capabilities will experience

large returns to the accumulation of additional capabilities. The model opens up two options

to the quiescence trap – increase in the average complexity of products, or increase in the

total number of capabilities that exist in the world. The trap calls for solving the coordination

problem between the accumulation of additional capabilities and the demand for those

capabilities.

Particular within Hausmann and Hidalgo (2010)’s empirically validated model,

calibration suggests that the world exists in a regime with a strong quiescence trap. More

so, when this model is combined with the results of Hausmann and Klinger (2006, 2007)

and Hidalgo et al. (2007), a more refined insight into the process of structural

transformation is revealed. This is “the ability to add a product to the production set of

a country depends not only on how close a given product is to an already existing one, but

also on how many other capabilities are present in the country and used in other, potentially

more distant, products” (Hausmann and Hidalgo, 2010: p. 27).

Furthering economic complexity analysis, Hidalgo (2009) finds that during the

42-year period 1963-2005 (while the product space remains relatively stable) only a few

highly dynamic economies have been able to considerably transform their productive

structures. Products, such as vehicles and machinery, populated the more densely



8

connected part of the network, while oil and some of its derivate products are always

located in a weakly connected periphery of the network. This demonstrates that oil requires

specific capabilities that do not foster development, despite generating large revenues.

Agricultural products and raw materials are also consistently located in the periphery of the

space. The countries that have transformed their productive structures most dramaticallyare Brazil, Indonesia, Turkey, Malaysia, Thailand, Republic of Korea, Singapore and China.

They have followed different trajectories. While the economic complexity of Republic of

Korea, Singapore and China were relatively high at the beginning of the observation period,

Brazil, Indonesia and Turkey started from a primitive production structure. Hidalgo (2009)

argues that in complex economies good governance and institutions may be all that is

required to stimulate capability building and economic growth. In turn, governments of

countries with low economic complexity should actively help catalyze market activities and

solve coordination problems associated with attempts to accumulate capabilities.

Freire (2011) creates an index of productive capacity of an economy using a variant

of the Methods of Reflections. This shows that it is very difficult for countries to improve

their productive capacities when they start from lower levels. Focusing on the period

1984-2009, he also finds that while there has been convergence in productive capacity

among countries that were initially above the average (relative to the world), those that had

below average productive capacity two decades ago have lagged further behind. This

suggests increasing overall divergence. The countries that have transformed the most in

the 25-year period are China, India and the United Arab Emirates. In the Asia-Pacific

region, Turkey, the Republic of Korea and Thailand are among the top 10 countries which

increased their productive capacity in relation to the average in that period (Freire, 2011).

With the exception of Bangladesh, all other least developed countries (LDCs) in the region

have ended the period further away from the world’s average in 2009 when compared with

their position in 1984 (Freire, 2011). LDCs of the region have productive capacity way

below the world average. Nepal’s productivity capacity, relative to the world average in

2009, was only marginally worse than in 1984. It is greater than that of all Asia-Pacific

LDCs, except Bangladesh, but lower than that of the developing countries in South Asia,

namely India, Pakistan and Sri Lanka. Regionally, it is average for Latin America and the

Asia-Pacific.

Using methods developed by Freire (2011) in a related research, ESCAP (2011)

finds that the countries that export the greatest number of categories of products and those

which have more products at different prices within those categories tend to have higher

levels of GDP. Findings also show that diversifications within and between categories of

products are not mutually exclusive. Richer countries continue to diversify, with the

dominant form of diversification being the expansion of production of different varieties

within the same category, as product categories rise. With average diversification of

countries increasing, but product mix becoming more standard, countries that do not

diversify are likely to fall behind (ESCAP, 2011).

Notably, only four countries – Estonia, Latvia, Lithuania and Vietnam – have

succeeded in transforming themselves during 1984-2009. The group began with

productivity capacities similar to those of the LDCs, and then raised it to above the world



9

average (ESCAP, 2011; Freire, 2011). The process of transformation was gradual, e.g. as

in Vietnam (ESCAP, 2011).

Based on the experiences of the countries that have transformed themselves,

ESCAP (2011) suggests a strategy for increasing productive capacities; this is made up of

three main processes for discovering, acquiring and spreading the productive capacities

required for developing economies to catch up. The first process is differentiation through

strategic product innovation, which is identification and production of products that are new

to the economy and more complex, facilitating further diversification. The second process is

through the selection of the business models of firms and farms that are successful in the

differentiation process. The third is the amplification of the successful business models and

the exploitation of the new market (ESCAP, 2011). These processes have to be repeated

continuously for the strategy to succeed (ibid.).

Regarding the first process, ESCAP (2011) finds that Asia-Pacific LDCs, onaverage, could product around 400 new products closely related to existing ones. However,

only 10-15 per cent of these would be both more complex and better connected to other

products. In the case of Nepal, there are around 514 products related to those already

exported, of which 58 (11 per cent), are more complex and better positioned for future

diversification. ESCAP (2011) suggests a pragmatic way to look for potential new products;

this is to emulate the production pattern of countries that have higher productive capacities,

even if they do not have higher per capita GDPs.

In the next section we review Nepal’s recent growth and export performance, usingsome of the product classifications developed in this paper.

B. Nepal’s economic and export performance:A preliminary analysis

In the 30-year period 1980-2010, Nepal’s GDP per capita growth has been poor

and erratic (figure 2). Annual compound growth of GDP per capita in the three decades was

a meagre 2.16 per cent, the lowest in South Asia (figure 3). This region is also home to

three other LDCs besides Nepal – Afghanistan, Bangladesh and Bhutan. While the annual

growth rate of GDP per capita for other countries in the region improved during the recent

decade (2000-2010), Nepal’s growth rate declinedto an average of 1.8 per cent per annum.

The contrast is even starker when compared with the GDP per capita growth rates of two

LDCs of Southeast Asia – Cambodia and Lao People’s Democratic Republic. These grew

at average annual rates of 6.5 per cent and 5.4 per cent respectively.

Nepal’s dismal GDP per capita growth has led to its per capita GDP to being the

lowest among South Asian countries, after Afganistan. This is also lower than the two

Southeast Asian LDCs – Cambodia and Lao People’s Democratic Republic (figure 4). Thegap with all these countries has widened since 1981. Notably, Bangladesh and Cambodia,

with about the same per capita GDP as Nepal’s in 1981 and 1995, had their per capita

GDP levels higher than Nepal’s by 38 per cent and 80 per cent respectively, by 2010.



10

5 500

5 000

4 500

4 000

3 500

3 000

2 500

2 000

1 500

1 000

500

0

1981 1985 1990 1995 2000 2005 2010

Afghanistan

Bangladesh

Bhutan

India

Nepal

Pakistan

Sri Lanka

Maldives

Cambodia

Lao PDR

P P P $ , c o n s

t a n t 2 0 0 5

Figure 2. Nepal per capita GDP growth performance (per cent)

Figure 3. Nepal in comparative perspective: Growth of GDP per capita

Source: Author’s calculation based on World Development Indicators.

Note: 2008 figure for Afghanistan’s GDP for 2010.


8

6

4

2

0

-2

-4

-6

1 9 8 8

1 9 9 0

1 9 9 2

1 9 9 4

1 9 9 6

1 9 9 8

2 0 0 0

2 0 0 4

2 0 0 6

2 0 0 8

2 0 1 0

1 9 8 0

1 9 8 2

1 9 8 4

1 9 8 6

2 0 0 2



11

Figure 4. Nepal in comparative perspective: Manufacturing, value added

(per cent of GDP), 2009


Note: Afghanistan’s figure is for 2008.

5 Excluding construction, and electricity, gas and water. Adding construction takes the ratio to over 58 per

cent.

6 Data on share of agriculture in GDP in 2010/11 and data on services are from Nepal Rastra Bank, “Recent

Macroeconomic Situation”, various issues.

The share of manufacturing value added in Nepal’s GDP was 6.79 per cent in 2009,

among the lowest in the set of comparator countries (figure 4), and also lower than for

LDCs as a group. During 1980-2009, Nepal’s manufacturing value added never reached

10 per cent of GDP (figure 5). However, while its share in GDP improved in the 1980s and

the early 1990s, it fell continuously after 2000. The share of agriculture in its GDP declinedmore sharply, going from 61.7 per cent in 1980 to 33 per cent in 2011. However, agriculture

continues to employ as much as 74 per cent of the currently employed labour force (CBS,

2008). Services5 (mostly of non-tradable variety) grew faster than agriculture and

manufacturing at 4.5 per cent per annum (compounded) during 2000/01-2010/11,

accounting for 52 per cent of GDP in 2010/11.6 The largest services sectors (in 2010/11)

are wholesale and retail trade (25 per cent), transport, storage and communication (19 per

cent), real estate, renting and business (16 per cent), education (13 per cent), and financial

intermediation (8.6 per cent). The structure of the Nepali economy has thus shifted from an

agriculture-dominated one to that of a non-tradable services-dominated one, with themanufacturing sector faltering.

20

18

16

14

12

10

8

6

4

2

0

A f g h a

n i s t a n

B a n g l a d

e s h

B h u t

a n I n

d i a

C a m b o

d i a

S r i L a

n k a

M a l d

i v e s

P a k i s

t a n N e

p a l

L a o P D

R

6.79



12

Figure 5. Share of manufacturing in Nepal’s GDP over time (per cent of GDP)


Nepal’s export performance, while relatively encouraging in the mid-to-late 1990s,

has been weak after 1999. Goods and services exports fell from 22.8 per cent of GDP in

1999 to 15.3 per cent in 2009. With imports increasing rapidly, fuelled and financed by

remittance earnings, the trade deficit (goods and services) has been burgeoning since theturn of the millennium, touching 21 per cent of GDP in 2009 (figure 6). Both merchandise

exports and services exports have performed poorly. However, services exports have

begun to recover after severely suffering between 1997-2007 (figure 7).

Figure 6. Nepal’s trade performance


12

10

8

6

4

2

0

1 9 8 0

1 9 8 2

1 9 8 4

1 9 8 6

1 9 8 8

1 9 9 0

1 9 9 2

1 9 9 4

1 9 9 6

1 9 9 8

2 0 0 0

2 0 0 2

2 0 0 4

2 0 0 6

2 0 0 8

1 9 8 0

1 9 8 1

1 9 8 2

1 9 8 3

1 9 8 4

1 9 8 5

1 9 8 6

1 9 8 7

1 9 8 8

1 9 8 9

1 9 9 0

1 9 9 1

1 9 9 2

1 9 9 3

1 9 9 4

1 9 9 5

1 9 9 6

1 9 9 7

1 9 9 8

1 9 9 9

2 0 0 0

2 0 0 1

2 0 0 2

2 0 0 3

2 0 0 4

2 0 0 5

2 0 0 6

2 0 0 7

2 0 0 8

2 0 0 9

30

25

20

15

10

5

0

-5

-10

-15

-20

-25

Exports of goods andservices (% of GDP)

External balance on goodsand services (% of GDP)

P e r c e n t o f G D P



13

Figure 7. Nepal’s exports of goods and services


Nepal’s exports-to-GDP ratio for 2009 is the lowest among the set of comparator

countries after Pakistan. Its trade balance-to-GDP ratio is among the worst (table 1).

Table 1. Nepal in comparative perspective: Trade performance

Exports of goods and services External balance on goods

(per cent of GDP), 2009 and services (per cent of GDP),

2009

Afghanistan 15.57 -32.1

Bangladesh 19.43 -7.1

Bhutan 58.55 9.8India 19.58 -4.4

Nepal 15.26 -21.1

Pakistan 12.84 -7.5

Sri Lanka 21.32 -6.5

Maldives 62.84 -23.4

Cambodia 59.61 -3.0

Lao PDR 28.94 -12.9


Note: 2008 figures for Afghanistan.

1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010

Service exports(BoP, current USD million)

Goods exports(BoP, current USD million)

1 200

1 000

800

600

400

200

0

U S D m i l l i o n



14

Not only has Nepal’s export performance been poor (in terms of value and growth),

its merchandise export basket remains low in technological sophistication and it is poorly

diversified with almost no change during the past decade.7 In 2010, Nepal exported 118

products with export value of more than $100, 000 (out of 277 possible products at 3-digit

level of SITC Rev. 3). In comparison, Bangladesh exported 213 and Cambodia 137.

8

Thetop 20 products made up 72 per cent of Nepal’s total merchandise exports.9

Figure 8 shows the evolution of the export basket since 1974/75 in terms of the

shares of nine SITC 1-digit categories. While manufacture products account for almost

70 per cent of exports in 2009/10, the share of food and live animals, and crude materials

and inedibles has increased in recent years. Although, this is still below levels reached in

Figure 8. Nepal’s export composition (in terms SITC 1-digit categories)

Source: Author’s calculation based on data from Ministry of Finance, Government of Nepal, Economic Survey, various issues.

7 Based on data from UNCTADstat.

8 Data from UNCTADstat.

9 Data from UNCOMTRADE.

2009/10

2008/09

2007/08

2006/07

2005/06

2004/05

1999/00

1994/95

1989/90

1984/85

1979/80

1974/75

Y e a r

0 20 40 60 80 100

Food & Live Animals

Tobacco & Beverage

Crude Materials & Inedibles

Mineral Fuels & Lubricants

Animals & Vegetable Oil & Fats

Chemicals & Drugs

Manufactured goods classifiedmainly by materials

Machinery and transportequipment

Miscellaneous manufacturedarticles

Commodity and transaction notclassified according to kind

Per cent



15

the mid-1980s. Animal and vegetable oils and fats, as high as 6.4 per cent in 1999/2000,

have a negligible share in 2009/10.10

Figure 9 shows the evolution of Nepal’s export basket in terms of the shares of

factor-intensity-based Leamer (1984) groups formed by aggregating SITC 2-digit

categories. The share of labour-intensive manufacturing has declined notably after 2003,

while the shares of capital-intensive manufacturing and tropical agriculture have increased.

Capital-intensive manufacturing made up almost half of total exports in 2010, while tropical

agriculture exports exceeded labour-intensive manufacturing exports. The sharp fall in

apparel exports since 2003 largely explains the declining share of labour-intensive

manufacture exports.11

10 The major destination of Nepal’s vegetable ghee was India. From 2002, India imposed quantitative

restrictions and non-tariff barriers on Nepal’s vegetable ghee exports. In subsequent years, India reduced and

eliminated ad valorem duty on imports of palm oil, the basic raw material used in the production of vegetable

ghee, eroding the competitiveness of Nepali exports, which was largely derived from the difference in the tariff

on the raw material.

11 Exports in HS Chapters 61 and 62 declined by almost 69 per cent during 2003-2010. The share of these

products in Nepal’s total exports declined from 34 per cent to 8 per cent in the same period (Based on author’s

calculation using UNCOMTRADE data). See Belbase and Kharel (2009) for the impact of expiry of the

Agreement on Textiles and Clothing on Nepali readymade garments export sector.

Figure 9. Nepal’s export composition in terms of Leamer categories

Source: Author’s calculation based on UN Comtrade, SITC Rev. 2.

2009

Y e a r

Petroleum

2003

2000

1995

1992

1988

1985

1982

0 20 40 60 80 100

Raw Materials

Forest Products

Tropical Agriculture

Animal Products

Cereals, etc.

Labour Intensive

Capital Intensive

Machinery

Chemical

2010

Per cent



16

The growth of iron and steel (SITC 67) and articles of metals, mostly of iron and

steel (SITC 69), contributed to the increase in the share of capital-intensive goods. Their

shares increased from 5 per cent and 2 per cent, respectively, in 2003 to 14 per cent and

3 per cent in 2010. Also contributing are textiles exports, whose share rose to 31 per cent

during this period.

12

The top 10 capital-intensive products (SITC Rev. 2, 4-digit) in 2010, interms of export value, accounted for 44 per cent of total exports and 89 per cent of total

capital-intensive goods exports. Besides iron and steel, these products included carpets,

synthetic yarn and fabric, jute twine, cordage and rope, among others. While carpet exports

fell, other textile exports increased in both quantum and share.

However, the high share of capital-intensive goods in Nepal’s export basket should

be interpreted with caution. In the Nepali context, some goods, classified with the Leamer

(1984) classification, are in practice relatively labour-intensive. For example, carpet exports

are mostly of the hand-knotted variety, and hand-knitting as well as use of hand-looms and

hand-and-footdriven machines, are still prevalent in production of other woolen goods. On

the other hand, while iron and steel products do represent capital-intensive modes of

production, it should be noted that domestic value addition (as a share of industry output) is

limited. This is due to complete dependence on imports for manufactured steel, and limited

transformation domestically (to be discussed later).

The share of raw materials has also increased substantially in the new millennium.

Data for 2010 shows that Nepal’s export basket has the highest capital-intensive

manufacture share and the lowest labour-intensive manufacture share among the export

baskets of the comparator countries, excluding Afghanistan, Bhutan and the Maldives(figure 10). In terms of importance of tropical agriculture, Nepal is second only to Sri Lanka.

Figure 11 shows the shifts in major export items between 2003 and 2010. Note the

declining share of apparel, and the rising shares of iron and steel and textiles.

Looking at changes in the composition of Nepal’s exports, in the decade from 1985

to 1995 the shares of primary products and resource-based manufactured products

decreased substantially. Meanwhile the share of low-technology products, comprising of

textile, garment and footwear (LT1), increased significantly (to over 85 per cent). This is

based on Lall (2000) classification of technological sophistication13 (figure 12). The shares

of other low-technology products (LT2) and medium and high-technology products were

negligible. However, by 2009 and 2010, the share of primary products and agro-based

manufactures had more than doubled. Also, the share of LT1 products dipped to 35 per

cent, the share of LT2 increased to 19 per cent, and the share of process-related medium-

technology (MT2) products increased to 13 per cent.

12 Iron and steel exports were negligible in 1995.

13 The Lall (2000) classification does not cover all SITC (Rev. 2) 3-digit codes. Specifically, it does not cover

animals, live, n.e.s., including zoo-anima, cinematograph film, exposed-develop, coin (other than gold) not

being leg, electric current, gold, non-monetary, printed matter, works of art, collectors pieces & an. Cambodia’s

export composition, in terms of Lall (2000) classification as shown in Figure 13, should be interpreted

cautiously because of the 32 per cent share of “unclassified” products.



17

Nepal

Pakistan

Maldives

Sri Lanka

Cambodia

India

Bhutan

Bangladesh

Afghanistan

0 20 40 60 80 100

Petroleum

Raw Materials

Forest Products

Tropical Agriculture

Animal Products

Cereals, etc.

Labour Intensive

Capital Intensive

Machinery

Chemical

Per cent

Figure 10. Comparative export composition – Leamer categories 2010


Note: 2010 data for all countries, except Bangladesh (2007).

Figure 11. Major exports

Source: WTO (2012). Trade Policy Review of Nepal.



18

100

Primary Products (PP)

90

80

70

60

50

40

30

20

10

0

1985 1995 2000 2009 2010

RB 2: Other

LT 2: Other Products

MT 2: Process

HT 1: Electronic and Electrical

Unclassified

RB 1: Agro-Based

LT 1: Textile, Garment and Footwear

MT 1: Automotive

MT 3: Engineering

HT 2: Other

Per cent

Figure 12. Nepal’s export composition – Lall classification

Source: Author’s calculation based on UN Comtrade, SITC Rev. 2, 3-digit level, and usingthe classification of Lall (2000).

The major LT2 products are iron and steel, which are heavily dependent onimported

intermediate goods (as noted above). Meanwhile, the shares of engineering-related

medium-technology products and high-technology products are negligible. The increase in

the share of primary products (to 22 per cent in 2010) corresponds to the decrease in the

share of manufacturing in GDP in the new millennium. Data for 2010 shows that Nepal’s

export basket has the highest LT2 and MT2 shares and the lowest LT1 share among the

export baskets of the comparator countries (figure 13), excluding Afghanistan, Bhutan and

the Maldives. In terms of importance of primary products, Nepal is second only to Sri

Lanka. In a sign of increasing commodity dependence, while the share of primary products

more than doubled during the decade, the share of resource-based manufactures declined,

albeit slightly.



19

C. Structural change and productivity growthin Nepal’s economy

When economic development and structural transformation takes place, overall

productivity growth occurs from within sectors (sectoral productivity growth), as well as from

shift in resources from low-productivity sectors to high-productivity sectors. In the case of

labour productivity, this implies that there are two components to productivity growth; this is

sectoral labour productivity growth and productivity growth resulting from reallocation of

labour across sectors, that is, changes in employment shares across sectors (McMillan and

Rodrik 2011). When employment share changes are positively correlated with productivitylevels, structural change contributes positively to productivity growth ( ibid.). In this section,

the McMillan and Rodrik (2011) approach is followed. Nepal’s labour productivity growth is

decomposed into these two components using sectoral value added and employment data.

Figure 13. Nepal in comparative perspective: Export composition

– Lall classification, 2010

Source: Author’s calculation based on UN Comtrade, SITC Rev. 2, 3-digit level, and usingthe classification of Lall (2000).

100

Per cent

90

80

70

60

50

40

30

20

10

0

A f g h a

n i s t a

n

B a n g l a d

e s h

B h u t

a n I n

d i a

C a m b o

d i a

S r i L a

n k a

M a l d

i v e s

P a k i s

t a n

N e p a

l

Primary Products (PP)

RB 2: Other

LT 2: Other Products

MT 2: Process

HT 1: Electronic and Electrical

Unclassified

RB 1: Agro-Based

LT 1: Textile, Garment and Footwear

MT 1: Automotive

MT 3: Engineering

HT 2: Other



20

This is to assess the nature and direction of structural transformation in the economy as

a whole.14

In the nine-year period of 1999-2008, labour productivity grew by a compound

annual growth rate of 1.44 per cent. The contribution of the “within” component (productivity

growth within sectors) was 0.88 percentage point (61 per cent). This is while the

contribution of the structural change component (the productivity growth resulting from

reallocations of labour across sectors) was 0.57 percentage point (39 per cent) (table 2).

That the contribution of structural change is positive is in line with McMillan and Rodrik

(2011)’s finding – that the aggregate structural change has been growth-enhancing in Asia.

However, in Africa and Latin America it has been growth-reducing, although the period

considered is longer – 1990-2005.15

Table 2. Decomposition of Nepal’s labour productivity growth during1999-2008 (average annual compounded growth rate)

per cent

Labour productivity growth 1.44

Within component 0.88

Structural change component 0.57

Source: Author’s calculations.

The positive contribution of the structural change component is also reflected in the

positive correlation between the log of the ratio of sectoral productivity to total productivity

at the end of the period in consideration (2008), as well as the change in sectoral

employment shares during the period (1999-2008), as depicted in figure 14.

14 Labour productivity is real value added divided by people employed in the sector. The decomposition is

done with the formula used by McMillan and Rodrik (2011):

∆Y t =Σ θ

i, t – k ∆ y

i, t +Σ y

i, t ∆θ

i, t

where Y t amd y

i, t refer to economy-wide and sectoral labour productivity levels at time t, θ

i, tis the share of

employment in sector i in time t, and the ∆ operator denotes change in productivity or employment shares

between time t-k and t.

The choice of the period of analysis is 1999-2008, which is determined by the availability of reliable

employment data disaggregated by major sectors. The chosen period corresponds to two points in time (fiscal

years 1998/99 and 2007/08) when labour force surveys were conducted in Nepal (CBS 1999 and 2008).

Employment is that of “currently employed” labour force, comprising people aged 15 years and above (CBS

1999 and 2008). Nine sectors are considered (See Table 3).

Data on value added by sectors are drawn from the Economic Survey , Ministry of Finance, Government of

Nepal, various issues. As the real value added data for 1998/99 are available at 1994/95 prices, the real valueadded data for 2007/08 are available at 2000/01 prices. The latter are rebased to 1994/95 prices by deriving

the ratio between the price deflators for the two base years for each sector from the ratio of real value added

data for 2001/02, at 1994/95 and 2000/01 prices.

15 They do not include Nepal in their sample.

i = n i = n



21

Notably, the somewhat surprising find in this paper is that Nepal’s relative

contribution of the structural change component to productivity growth is two and a half

times that for the un-weighted average for Asia (14.7 per cent) in the period 1990-2005.

However, in absolute terms the total annual labour productivity growth is quite low, at only

1.4 per cent. Productivity growth is negative in five of the nine sectors, including

manufacturing. Low overall productivity growth is largely due to the low “within” sector

productivity growth, contributing less than 1 percentage point. In contrast, as shown in

McMillan and Rodrik (2011), general “within” sector productivity growth is much higher in

high-growth countries, as well as the key driver of total productivity growth. When “within”

productivity growth is low or negative and productivity levels are also low, there is only so

much that the structural change component can contribute. The latter can only supplement

growth “within” sectors under such circumstances.

The manufacturing sector is a key tradable sector with high potential to absorb

labour. It has the lowest productivity level after agriculture, and has exhibited negative

productivity growth as well as low total value added growth. Its employment share has

increased by only 0.7 percentage points during 1999-2008 (table 3). It is the only sector

(besides mining – of minor significance in the economy) to have witnessed a decline in the

share in total value added (by 2.3 percentage points), despite registering a small increase

in employment share. The largest increase in employment share has been in the

wholesale, retail trade, restaurants and hotels sector, which is largely a non-tradable sector

Figure 14. Correlation between sectoral productivity and change in employment

shares in Nepal (1999-2008)

Source: Author’s calculation.

Change in employment share (1999-2008)

Log of sectoral productivity/total productivity (2008) Fitted values

-.02 -.01 0 .01 .02

-1

0

1

2

3

L o g o f s e c t o r

a l p r o d u c t i v i t y / t o t a l p r o d u c t i v i t y ( 2 0 0 8 )

agr

csp gs

con

tsc

fin

egs

mfg

min

wrt



22

(except for its tourism component).16 The community, personal, social and government

services sector – another non-tradable services sector – has recorded the highest

productivity growth rate. Agriculture, despite being the least productive sector, still employs

about 74 per cent of the labour force. The employment share of agriculture fell by only

2.2 percentage points.

Table 3. Sectoral employment, value added shares and their change, and

productivity level and growth (Nepal)

Employment Value Productivity Change in Change in Productivity

share added 2008 employment value growth (%,

2008 share (NPR, share added compound

2008 constant (1999- share annual,

1994/95 2008) (1999- 1999-

prices) 2008) 2008)

Agriculture, 0.74 0.37 15,231.19 -0.02 -0.01 1.47

forestry, fishing

Mining and 0.00 0.01 73,969.93 0.00 0.00 -9.24

quarrying

Manufacturing 0.07 0.08 35,018.69 0.01 -0.02 -2.74

Electricity, gas 0.01 0.02 69,840.27 0.01 0.01 -7.09

and water

Construction 0.03 0.11 106,421.83 -0.01 0.00 2.71

Wholesale, 0.08 0.09 35,137.95 0.02 -0.03 -4.99

retail trade, hotels

and restaurants

Transportation, 0.02 0.08 151,632.23 0.00 0.01 1.22

communication,

storage

Finance and 0.01 0.12 422,924.29 0.00 0.02 -1.94

real estate

Community, 0.05 0.12 71,379.13 -0.01 0.03 7.13

social, personal

and government

services

All sectors 30,270.74 1.44

Source: Author’s calculations.

Note: Zero values in shares and changes in shares denote extremely small values.

16 Sub-sectoral data on value added in this sector are not available for 1998/99. But they are available for

2007/08. In that year, wholesale and retail trade value added was 7.6 times higher than restaurants and hotels

value added.



23

There is a massive under-utilization of labour. Conservative estimates puts the

labour under-utilization rate at 30 per cent in 2008, although the official unemployment rate

is just 2.1 per cent (CBS, 2008).17 Furthermore, the above analysis of labour reallocation

across sectors within the domestic economy, is based on the active labour force, hence

excludes Nepali nationals working abroad. Lack of productive employment opportunities inthe country is driving people to migrate abroad for work in increasing numbers. There has

been an annual outflow of over 200,000 people to destinations other than India in recent

years (or about 50 per cent of the annual addition to the labour force). The total number of

migrants abroad is at least 2 million, with a quarter of the adult male population working

abroad.18 Therefore, although reallocation of employed labour in the aggregate has been

from low-productivity to high-productivity sectors in the last decade, faster transfer of labour

from agriculture to more productive sectors within the economy is constrained by a weak

manufacturing sector. This sector is experiencing declining productivity, and whose

productivity level relative to total productivity is lower than for the world on average.

19

Here, the emphasis on manufacturing should not be interpreted as ignoring the

potential of the modern services sector, such as tourism. The emphasis is because:

i) manufacturing constitutes a tradable sector and global experience shows that it combines

relatively high productivity with the potential to absorb low-skilled labour from agriculture;

ii) there is empirical evidence that, unlike economies as a whole, manufacturing industries

exhibit unconditional convergence in labour productivity across economies (see Rodrik

2011a, 2011b); iii) development of tradable sectors is critical for reducing Nepal’s excessive

reliance on remittances for foreign exchange; iv) while tourism is a key tradable service

sector with high potential for employment generation and it should no doubt be developed,

arguably a sustainable growth strategy for a small open economy calls for robust expansion

of bothtradable services and manufacturing sectors; v) the poor performance of Nepal’s

manufacturing sector absolutely calls for urgent attention. Notably, it is relative to other

sectors of the Nepali economy, as well as relative to other countries/groups/regions (e.g.

South Asia, East Asia, LDCs). Productivity in manufacturing is even lower than that of the

community, social, personal and government services sector, which is the least productive

sector in the world (see McMillan and Rodrik, 2011). This low productivity and extremely

slow expansion of manufacturing suggests that constraints to growth have been especially

severe for the sector. The poor performance of the manufacturing sector is also reflected in

the poor performance of merchandise exports, in terms of both earnings and export basket

composition.

17 Comparable under-utilization data is not available for 1999.

18 Stock and flow employment data compiled from CBS (2008), CBS (2011); CBS (2012), Economic Survey ,Government of Nepal, Ministry of Finance, various issues and “Nepal Migration Survey”. Preliminary findings

are presented by the World Bank, August 2011.

19 The comparison is with the average global figures for manufacturing productivity and total productivity for

2005 presented by McMillan and Rodrik (2011).



24

D. Nepal’s trade and industrial policies in the contextof structural transformation

Nepal introduced a new Trade Policy in 2009, replacing the Trade Policy 1992. The

main objective of Trade Policy 2009 is to support economic development and povertyalleviation initiatives through “the enhanced contribution of trade sector to the national

economy” (GoN, 2009). Its specific objectives are:

• To create a conducive environment for the promotion of trade and business in

order to make it competitive at the international level

• To minimize the trade deficit by increasing exports of value-added products

through linkages between import and export trade

• To increase income and employment opportuni ties by increasing

competitiveness of trade in goods and services, and using that as a means of

poverty alleviation

• To clearly establish the interrelationship between internal and foreign trade,

and develop them as complementary and supplementary to each other

Promoting exports, in order to facilitate structural transformation of the economy, is

not an explicit aim of the trade policy. The emphasis on exports of “value-added” products

is motivated more by the need to reduce the trade deficit. However, fostering of backward-

forward linkages, product diversification, and domestic value addition and processing do

feature in strategies, and specific and working policies. This includes those at the sectoral/

product level.

Trade Policy 2009 aims to promote the exports of both goods and services.

However, it discusses goods trade elaborately, envisaging the promotion and development

of select goods, and classifying them into “Special Focus Area” and “Special Thrust Area”.

Whereas, Trade Policy 1992 did not target any products, as such. In the first group are the

labour-intensive goods already established in export markets, such as readymade

garments, woolen carpets and handicraft goods. The second group includes “highly

potential export items” (GoN, 2009: p. 29), mostly agricultural and forest-based products.The rationale for promoting exports of agricultural and forest-based products is based on

the Agricultural Perspective Plan and Periodic Plan accepting agriculture as a priority sector

of the economy. Recognition of the important role of agricultural development in poverty

reduction is also part of this rationale. Table 4 lists the identified goods. The “Special Focus

Area” has four goods, while the “Special Thrust Area” has 15 goods. There is some overlap

between the Special Focus Area and the Special Thrust Area.



25

Rationale for selection of goods for export promotion is “goods that are of

comparative advantage, and based on skills, means and resources available in the country

will be identified and selected, and promotion of trade in such goods will be supported.”

(GoN, 2009: p. 4). Utilizing local resources to create employment and income generating

opportunities in rural areas is the main rationale for the dominance of agricultural and

forest-based products in the Special Thrust Area. Trade Policy 2009 also provides for the

identification of new exportable goods in which the country has comparative advantage.

Emphasis is on commercial farming, livestock, and non-timber forest products.

Policies for the Special Thrust Area goods stress value addition, processing,

forward linkages (using primary goods for the production of manufactured goods) and

vertical product diversification. Even within the Special Focus Area, which only includes

manufactured goods, there is an emphasis on backward linkages. This is done by

increasing the domestic production of raw materials and inputs, including agriculture and

forest products. There are also policies for value addition and vertical diversification for

manufactured products in the identified Special Focus Area. However, concrete valueaddition and product diversification strategies as well as programs for turning commodities

into manufactured goods are lacking. This is important in order to avoid the possible trap of

commodity dependence.

Table 4. Products identified in Trade Policy 2009

Special focus area Special thrust area

1. Readymade garments, 1. Tea 9. Herbs and essential oils

cotton towels

2. Woolen carpets 2. Vegetable seeds 10. Handmade paper and

paper products

3. Pashmina and silk 3. Large cardamom 11. Wood craft products

products

4. Handicraft goods 4. Pulses 12. Coffee

(Pashmina, woolen

products, silver products,

metal products,

handmade paper)

5. Floriculture 13. Honey

6. Precious/semi-precious 14. Oranges ( junar )

gems and stones,

and gold and silver

ornaments

7. Processed leather 15. Vegetables

(and leather goods)

8. Ginger/dried ginger Source: Author’s compilation based on Trade Policy 2009 (GoN, 2009).



26

The Trade Policy 2009 strategy of mostly promoting goods that intensively utilize

domestically available resources has merit in terms of addressing immediate concerns of

unemployment/underemployment of human resources and poverty. However, it does not

address the question of what potential the production and export of the selected goods

holds for structural transformation. This is needed for rapid and sustained growth. In other words, still to be investigated is the potential of identified goods for upgrading the national

industrial production structure. This would enable the economy to progressively produce

more sophisticated and complex products. Importantly, (as discussed in Section 3) recent

research consistently shows that the type of goods that a country currently exports

determines the type of goods it will export in future, and its future economic growth rate.

Following Trade Policy 2009, the Government of Nepal created the Nepal Trade

Integration Strategy (NTIS) 2010. This strategy identifies 19 “priority export potential

sectors” – goods and services – and the “most attractive markets” for them. Focusing on

the identified sectors, it charts a short- to medium-term course of action for the

development of the country’s export sector until 2013-2015 (see GoN, 2010a). Among the

19 sectors, 12 are goods (7 agriculture and food products, 5 craft and industrial goods) and

7 services (table 5). Most of them are also included in the Trade Policy 2009 special focus

and thrust areas, and have been in Nepal’s export basket for decades. NTIS 2010 has two

sectors/products that are not identified in the Trade Policy 2009. Whereas, the Trade Policy

2009 identified nine sectors/products that are not in NTIS 2010 (namelyready made

garments and cotton towels, woolen carpets, vegetable seeds, floriculture, processed

leather – and leather goods, wood craft products, coffee, oranges junar , and vegetables).

Among the 12 identified goods with export potential, NTIS proposes to focus, in the short-

to medium-term, on agriculture (including forest) and food products.

Table 5. Priority export potential products/sectors identified by Nepal Trade

Integration Strategy 2010

Products Whether the product is identified

in Trade Policy 2009 (only goods)

Agro-food1 Cardamom Yes

2 Ginger Yes

3 Honey Yes

4 Lentils Yes

5 Tea Yes

6 Noodles No

7 Medicinal herbs/essential oils Yes



27

The 19 goods and services sectors are identified based on an initial assessment of

export performance and some extensive discussions with Nepalese business community

and government officials. The sectors’ aim is diversifying exports product-wise and

destination-wise, moving up the value chain and ensuring that exports have a robust,

positive impact on inclusive growth (GoN, 2010a: p. 10). Most of the selected sectors are

identified as having medium or high export potential, or medium or high socio-economic

impact, or both (GoN, 2010a).

However, it must be noted that the assessment (using a number of indexes) of the

export potential of the 19 goods was done ex post , i.e. after they were identified. Thus, theexport potential of the identified goods or services is to be interpreted regarding one

another, not with respect to all goods or services. The factors, considered in the

construction of the export potential index used in NTIS 2010, are the current export

Craft and industrial goods

8 Handmade paper Yes

9 Silver jewelry Yes

10 Iron and steel No

11 Pashmina Yes

12 Wool products Yes

Services

13 Tourism

14 Labour services

15 IT and BPO services

16 Health services

17 Education

18 Engineering

19 Hydro-electricity

Other potential export sectors

20 Transit trade services

21 Sugar No

22 Cement No

23 Dairy products No

24 Transformers No

Source: Author’s compilation based on GoN (2009) and GoN (2010a).

Table 5. (continued)

Products Whether the product is identified

in Trade Policy 2009 (only goods)



28

performance of Nepal in a given product, current world demand and market access

conditions for the product, as well as domestic supply capacity. This last factor includes

quality of products, the productivity and cost of production factors, and the efficiency of

supporting domestic industries.

Likewise, the potential socio-economic impact is measured by a socio-economic

impact index. This is a composite indicator of full-time equivalent employment, participation

of women in the sector, impact on poor regions, and impact on skill development. The

resource intensity (defined as dependence on electricity and water) of most of these

sectors was found to be medium to low. This implies that the promotion of these sectors will

not be constrained by two current critical resource bottlenecks (GoN, 2010a). Together, the

19 sectors cover about 30 per cent of goods exports in 2008, and the vast majority of

service exports – broadly defined to include exports of labour services (GoN, 2010a). NTIS

2010 also identifies five “other” sectors (four goods and one service) with export potential

(table 5).

NTIS 2010 shows some recognition of the need for enhancing value addition,

processing and diversification of the identified products, including agro-food products that

are accorded high-priority. However, the methodology employed for product/sector

identification does not explicitly take into account the possibilities the selected products

hold (or do not hold) for future product diversification and enhanced export sophistication.

Trade is identified as one of the six pillars of development strategy for the

realization of the goals (poverty reduction) and objectives of the Government of Nepal’sThree-Year Development Plan 2010/11-2012/13 (GoN, 2011). This is the first period

development plan to focus on mainstreaming trade in order to achieve development

objectives. Export trade is one of the priority sectors in the Plan. Guided by the Trade

Policy, the Plan envisages reducing trade deficits by developing exportable goods and

services to having comparative advantage, as well as how best to utilize the opportunities

created by the bilateral, regional and multilateral trading systems. The objectives of the

Plan, with respect to trade, are to enhance income and employment opportunities by

promoting domestic and international trade; to derive maximum benefits from goods and

services trade by identifying comparatively new beneficial goods and services. This, withthe participation of the private sector and the government, is to make price and quality

more competitive in internal and external markets; to alleviate poverty by promoting exports

of goods using local raw materials, resources and skills; more so, to expand the benefits

from trade to rural areas (GoN, 2011).

One of its working policies is identification, development, promotion, marketing, and

establishment of value chains of new exportable goods and services. Another working

policy is to implement the NTIS 2010. The strategies include developing trade as an

important pillar of the national economy. This means increasing value addition in exportable

products, value chain development and identification, and promotion of new export

potential goods (both horizontal and vertical diversification). Also, in the strategies is the

mobilization of foreign aid to increase export competitiveness, and provide incentives,

facilities and concessions for exportable products. At the programme level, the Plan



29

emphasizes the development of exportable agriculture and forest-based products. Given

that the Plan is guided by the Trade Policy and intends to implement NTIS, the products

identified in the policy and the strategy is expected to be targeted during the Plan’s

implementation.

The Government introduced a new industrial policy in 2010. This is in response to

the poor performance of the industrial sector following the unsatisfactory implementation of

the Industrial Policy 1992. It is also due to new opportunities and challenges on the

industrial front (GoN, 2010b: 1-3). The long-term goal of the Industrial Policy 2010 is to

contribute to poverty alleviation through sustainable and broad-based industrial

development. A major objective is to increase the national income and employment. This

includes increasing exports of industrial goods through expansion of quality and

competitive industrial production, and rise in industrial productivity. The policy also aims to

increase the contribution of the industrial sector to national and regional development

through the mobilization of local resources, raw materials and skills. Its strategies, with

respect to the industrial export sector, emphasizes increasing domestic value addition in

exportable goods, fostering domestic forward and backward linkages, and development of

industries (including agro-based) utilizing domestically available raw materials .

The Industrial Policy 2010 provides facilities and concessions (including in taxes) to

export-oriented industries, among others. It also provides for additional facilities and

concessions, (including in taxes) to export-oriented industries located in special economic

zones – these are yet to be established. The Industrial Policy 2010 declares 10 industries

as priority industries entitled to additional facilities and incentives. The list covers a varietyof goods and services (e.g. particular tourism-related sectors, computer software), as well

as non-good/service-specific industries (e.g. export-oriented industry and traditional cottage

industry). Notably, the agriculture and forestry-based industry (listed as a priority industry)

includes both farming/cultivation (agriculture) and processing/manufacturing. The inclusion

of agriculture as an industry may be due to the critical importance of its production in

raw material requirements of agriculture and forestry-based industries.20 Relatively

sophisticated (non-services)21 priority industries include clinker and cement production

based on domestic limestone; pulp and paper production; chemical fertilizer production

(other than simple mixing); powder milk production; pharmaceutical production; productionof fuel-saving equipment; production of pollution-reducing equipment; production of

equipment and gadgets used by the physically challenged; production of agriculture tools

and equipment and industrial machinery; and production of electric vehicles. Among these,

only three products (in broad terms) feature in the trade policy and/or NTIS. Paper has

been identified for export promotion in both the trade policy and NTIS, and cement and

dairy products (which included powder milk) are among the “other” potential export sectors

in NTIS. Changes to the Industrial Enterprises Act are needed to incorporate some of the

provisions in Industrial Policy 2010.

20 A related issue is that the definition of “industry” by the policy is very broad as such, encompassing even

real estate business.

21 We consider energy production and distribution, including hydropower production and distribution, as

a service industry.



30

As noted in the previous section, there is a close link between the sophistication,

complexity and diversity of products produced by an economy, the structural transformation

and growth paths of the economy. Sophistication, complexity and diversity of an economy

are reflected in its export basket. The degree of export-orientation varies across industries,

as demand and competition conditions for different products differ in the domestic marketand foreign markets. The relatively sophisticated manufacturing products accorded priority

by the Industrial Policy may not be targeted for exports in the immediate future. But the fact

that they are accorded priority may reflect an aim to create and/or enhance the capabilities

of the economy to produce such products, at least for the domestic market. An economy’s

ability to produce more and more diverse and complex/sophisticated products is likely to

produce a similar effect in its export basket. Equally, increasing the diversity and

complexity/sophistication of exports is also likely to help make the industrial structure of the

economy more diverse and complex/sophisticated, as the exportable products are

produced within the economy. The potential synergy between the industrial policy and thetrade policy/NTIS for industrial upgrading and structural transformation remains to be

investigated.

Neither Trade Policy 2009 nor NTIS 2010 take into account the fact that all products

do not hold the same prospects for structural transformation and economic growth. In which

part of a product space a country’s exportable products are concentrated and what

products (upscale or downscale) the country exports also matter for future rates of income

growth. Given that the Government of Nepal has taken trade as an engine of growth in its

Plan document, it is important that this structural transformation dimension is brought into

trade policy making and implementation.

E. Methodology for assessing structural transformationthrough the “export” lens

1. Methods, tools and measures

Methods, tools and measures developed in the empirical literature discussion on

structural transformation, are employed to describe the pattern of evolution of Nepal’smerchandise exports. These will then assess the structural transformation-effecting

potential of the goods identified for export promotion in Nepal’s Trade Policy 2009 and the

NTIS 2010 (Hausmann et al., 2006, Hidalgo et al., 2007; Hausmann and Klinger, 2006,

2007; Felipe et al., 2010c; and Abdon and Felipe, 2011).

Measures for analysis of the following three variables are the level of productivity/

sophistication associated with a particular product; the level of sophistication of the export

basket of a country (or a set of products exported/targeted for exports by a country); the

inter-relatedness between and among products (or how close a product is to other products).

We use the measure of productivity associated with a product (PRODY) developed

by Hausmann et al. (2006). PRODY is a weighted average of the GDP per capita of the



31

countries that export the product. The weights are taken as the ratio of the revealed

comparative advantage (RCA) of each country in the product to the sum of the RCAs of the

all countries in the product:22

PRODYi = Σ j Yj , where xij is exports of country j of product i, Xj is total exports of

country j, and Yj is GDP per capita (PPP$) of country j.

We use the measure of sophistication of export basket of a country (EXPY),

developed by Hausmann et al. (2006). EXPY is the weighted average of the PRODY of

products exported by a country, with the weights being the share of the product in the

country’s total exports:

EXPYj =Σ ( ) PRODYi

We use PRODY and EXPY to measure the sophistication/complexity of products/

export basket/economy. These are used instead of the better measures of complexity

based on the Methods of Reflection developed by Hidalgo and Hausmann (2009), that only

considers network information and does not use income information. This is because

PRODY and EXPY are easier to construct and, more importantly, there is a strong

correspondence between PRODY and EXPY with their network counterparts as discovered

by Hidalgo (2009). This suggests that “most of the information contained in PRODY and

EXPY comes from the structure of the network connecting countries to the products they

export, rather than from income” (Hidalgo, 2009: p. 7). Thus, the power of PRODY andEXPY in explaining growth and structural transformation (e.g. Hausmann et al., 2006)

“comes from the information on the diversification of countries and on the ubiquity of

products” (Hidalgo, 2009: p. 8).

We use the outcome-based measure of proximity between two products (øij)

developed by Hausmann and Klinger (2006, 2007) and Hidalgo et al. (2007).

Mathematically, the proximity between two products i and j is defined as:

Øij = min {P (RCAi|RCAj), P (RCAj|RCAi)}.

Proximity between the two products i and j is, therefore, the minimum conditional

probabilities that a country exporting one good with comparative advantage (RCA>1) also

exports the other with comparative advantage.

These three basic measures are used to calculate four other measures: Path,

Density, Open Forest and Strategic Value.

xij / xj

Σ j xij / xj

i

xi

Xi

22 The RCA measure used by Hausmann et al. (2006) is different from the more popular RCA index of

Balassa (1965).



32

The Path associated with a product (i) is the measure of the inter-connectedness of

that product with all other products. It is calculated as: Path i = Σ j Øij (Hausmann and

Klinger, 2006). It indicates the potential for future export diversification associated with

product i.

Density associated with a product (i), developed by Hausmann and Klinger (2006,

2007), is a measure of the ease with which a country can deploy its existing capabilities to

produce it, that is, proximity of the product to the current export basket of the country.

Density measures the degree to which a country’s current exports “surround” the particular

product under consideration. It is the sum of all proximities between the product and all

products in which the country is present (has comparative advantage), scaled by the sum

of all proximities leading to the product. It varies from 0 to 1, with higher values indicating

that the country is more likely to effectively export that product in the future.

Mathematically, densityi = where xik = 1 if the country is present in the product,0 otherwise.

Open Forest represents the “option value” of a country’s unexploited opportunities,

an option set for future structural transformation (Hausmann and Klinger, 2006 and 2007). It

is the distance-weighted value of all the products a country could potentially produce,

where the distance is the relative distance of each product not currently effectively exported

to the current export basket. This is calculated as:

open forest =

Σ j (1 – xj ) PRODYj where Øij is proximity and

x i , x j = { 0 }; is the density

The Strategic Value of a product is a proxy for the spillovers derived from acquiring

comparative advantage in that product (Felipe et al., 2010c). It is the increase in the Open

Forest assuming that the country gains comparative advantage in that product (ibid.). The

Strategic Value of a product j is calculated as:

Vj = Σi (1 – xi ) PRODYi , for all i, i≠ j, xi=1 if RCA>1

Strictly speaking, the concept of Strategic Value pertains to products in the Open

Forest only, that is, those products not currently exported with comparative advantage.

However, we calculate Strategic Values of all identified products, even if they happen to not

be in the Open Forest (i.e. they are already being exported with comparative advantage in

the RCA>1 sense). The reason is provided later in the paper. In such calculations too, the

formula remains the same.

We also use the classification of products employed by Abdon and Felipe (2011), in

terms of distance from the current export basket. Abdon and Felipe (2011) describe

products as “nearby” if the distance (density) is less than 0.5 standard deviations from the

Σk øik xik Σk øik

Σi øij xi

Σ øij

if RCA > 11

otherwise

Σi øij xi

Σ øij

øij

Σ j øij



33

mean distance, “middle” if the distance is within ± .5 standard deviations from the mean

distance, and “far away” if the distance is more than 0.5 standard deviations from the mean

distance.

We also use the classification of products as high-path/PRODY, mid-path/PRODY

or low-path/PRODY depending on whether they belong to the first, second or third tercile of

path/PRODY, as used by Felipe et al. (2010c).

Using these methods, tools and measures, we first describe the evolution of Nepal’s

exports in the product space during the 15-year period 1995-2010. Three points in time are

considered – 1995, 2003 and 2010. Also described is how the sophistication of the

country’s export basket and the Open Forest associated with its export basket (indicating

the prospects for structural transformation) has evolved over the period. This sets the

context forassessing the potential held by the products identified by Trade Policy 2009 and

NTIS 2010 for future structural transformation.

2. Data

We use the values of PRODY and proximity (Øij) calculated by Hidalgo et al.

(2007). They are the two basic measures required for analysis, as well as being the basis

for calculation of EXPY, Open Forest, Paths and Density. The level of disaggregation of

products is the four-digit Standard International Trade Classification (SITC) Revision 2.

Their dataset covers 775 products. Proximity values are missing for about 5 per cent of the

possible product pairs. We consider them to be zero, that is, the products are unrelated.

The fact that the PRODY used by Hidalgo et al. (2007) is in constant 2000 PPP$

will not affect the study’s analysis when we compare values between cross-sectional units

(countries), or over time for a cross-sectional unit, or between cross-sectional units over

time. The proximity calculated by Hidalgo et al. (2007) is an average of the proximity for the

three years 1998-2000. The relatively dated data of proximity (and also PRODY) will not

substantially alter our results because Hidalgo (2009) finds that during the 42-year period

1963-2005 the structure of the product space (connectivity among products) remained

relatively stable. Also, the level of product sophistication remained relatively stable during

the 20-year period 1985-2005.

Although some papers have calculated PRODY and proximity values with more

recent data, and/or using a greater level of product disaggregation (for example, Felipe

et al., 2010a, 2010b, Abdon et al., 2010 and Hidalgo, 2011), we use the values calculated

by Hidalgo et al. (2007) because they are publicly available, while others are not.

The products identified in NTIS have been specified in Harmonized System (2002)

codes at the six-digit level. We convert them into SITC Rev. 2 4-digit classification. The

relevant SITC codes are allocated for the products identified in Trade Policy 2009 as theyare not specified in terms of international classification. Use of SITC classification instead

of HS classification entails a loss in product diversity. However, this is unavoidable given

the availability of PRODY and proximity values only at the SITC 4-digit level. Code



34

conversion is done using the concordance table available at World Integrated Trade

Solution (WITS).

Export data is taken from UNCOMTRADE, via WITS. Direct export data for Nepal

are available for every year in the period 1982-2000, after which data is only available for

2003, 2009 and 2010. The number of SITC 4-digit products exported by Nepal fluctuates

sharply during 1982-1995, with apparently inexplicable spikes every three years or so. We

take 1995 as the initial year when analysing the evolution of the export basket in detail.

3. Limitations

Our method and approach suffer from a number of limitations. The PRODY and

EXPY measures, based on gross trade flow data, do not fully take into account the type of

production activities that occur in the exporting country. In the presence of international

production fragmentation and intermediate good trade, a country’s exports does notnecessarily reflect the embodied technology and relative endowments that have gone into

the country’s domestic production activities (Aashe and Gangnes, 2007). Further, these

measures only consider across-product sophistication. It does not consider “within-product”

sophistication or product quality (see Xu, 2010).

The approach is based on exports, and does not consider production and industrial

capabilities not associated with exports, but which are nonetheless important. Only

merchandise exports are considered. Service exports, which are globally faster growing

than goods exports (and which are important for Nepal), are not considered because dataconstraints do not make services amenable to product space analysis.

In our discussion, we acknowledge these and other limitations. We emphasize,

where relevant, the need to interpret and analyse the results carefully, taking into account

the country-specific peculiarities and needs.

F. Analysis of export baskets and identified products

1. Analysis of export baskets over time

Using the PRODY values (2000 PPP$) calculated by Hidalgo et al. (2007), for

products under SITC Rev. 2 at the 4-digit level, we calculate and compare the

sophistication of Nepal’s export basket and that of the comparator countries.23

23 PRODY values calculated by Hidalgo et al. (2007) are missing for some products exported by the

countries under consideration. Products with missing PRODY account for less than 1.2 per cent of exports of

Pakistan, about 2 per cent of exports of Sri Lanka, about 5.4 per cent of exports of Cambodia, and negligible

percentage of the exports of other countries. While calculating EXPY, the weights are taken as in the originalformula (shares of products in total exports). This approach does not change EXPY values drastically, although

EXPY for Cambodia it is 10,317.64 instead of 9,675.14 when the exports of products with missing PRODY are

deducted from the total export figures while calculating the weight shares. But while calculating the distribution

across PRODY groups and across countries (Figure 16), a separate group for products with missing PRODY is

also created.



35

A time series plot of EXPY shows a downward trend depicted in export

sophistication during the mid-1980s. This failed to recover lost ground until the mid-1990s,

but exhibited a rising trend thereafter (figure 15).24 The rise in EXPY was rapid during

the period 1995-2003, with a compound annual growth rate (CAGR) of 13.8 per cent.

However, growth slowed down substantially thereafter, with the CAGR of EXPY for theperiod 2003-2010 barely 0.7 per cent. In 1995, 83 per cent of exports are in the low

PRODY, low Path category, while less than 4 per cent of exports are in the middle PRODY,

high Path category. By 2003, the share in the first group halved to 42 per cent and that in

the second group increased fivefold to 19 per cent. The shifts in the relative size of the two

groups are far less pronounced during 2003-2010. The first category’s share declined by

35 per cent and the second category’s share increased by 60 per cent. Products in the

so-called “core” of the product space (metals, machinery and chemicals, as defined by

Felipe, 2010a) made up 1.6 per cent of total exports in 1995, the share increased to

14.3 per cent in 2003, and at a slower rate to 24 per cent in 2010.

Figure 16 shows that Nepal’s export sophistication, as measured by EXPY, is higher

than that of Sri Lanka and Pakistan, and Afghanistan and Bangladesh. However, it is less

than that of India, Bhutan, Cambodia and the Maldives. Notably, the EXPY measure should

be interpreted with caution, as the extent of domestic value addition and transformation in

production is not reflected in the measure. It only considers the nature of the final good

exports (to be discussed in more detail later).

Figure 15. Evolution of EXPY


24 Data are missing for the years 2004-2008.

10 000

9 000

8 000

7 000

6 000

5 000

4 000

3 000

2 000

1 000

0

1 9 8 2

1 9 8 3

1 9

8 4

1 9

8 5

1 9 8 6

1 9

8 7

1 9

8 8

1 9 8 9

1 9 9 0

1 9

9 1

1 9 9 2

1 9 9 3

1 9 9 4

1 9 9 5

1 9 9 6

1 9 9 7

1 9 9 8

1 9 9 9

2 0 0 0

2 0 0 3

2 0 0 9

2 0

1 0

E X P Y ( 2 0 0 0 P P P $ )



36

Figure 16. Nepal in comparative perspective: Export sophistication 2010

Source: Author’s calculation based on UNCOMTRADE, SITC Rev. 2.


In figure 17, we divide the products exported by Nepal into three six PRODY groups

(in increasing order of PRODY range) and determine their shares in total exports at

different points in time. We find that the share of group 1 (PRODY<2000) has decreased

sharply since the 1990s to 10 per cent in 2010. Group 4 accounts for 36 per cent of

exports, the largest share. Group 5 has a 6 per cent share and Group 6 less than 0.2 per

cent. In 1982, total exports consisted almost entirely of Groups 1, 2 and 3. During

2003-2010, the share of Group 2 increased while that of Group 3 decreased. Figure 18

shows that after Afghanistan and Sri Lanka, Nepal has the highest share of Group 1

products in its export basket. However, at the same time, the country has the highest share

of Group 4 products in its export basket, after Bhutan. India, Cambodia and the Maldives

have a significantly higher share of Group 5 products in their export baskets than Nepal. At

1.5 per cent, the share of Group 6 products in India’s export basket is the highest among

the export baskets of the countries compared.

The above analysis indicates that Nepal’s export performance, in terms of export

earnings growth, has been poor since the late 1990s. However, in comparison to both its

past performance and to performance of comparator countries, the country fares relatively

better in terms of the sophistication of export basket. But sophistication of its export basket

has improved very slowly since 2003. The export sector has been unable to make inroads

into products in the higher range of technological sophistication. Also, the share of primary

products in the export basket has been increasing in the last decade. The challenge thatNepal faces on the export front is two-fold: a) to increase the export earnings by increasing

the export of existing products. Given the current industrial and export structures and

capabilities, this aims at turning around the deteriorating trade balance and increasing

12 000

10 000

8 000

6 000

4 000

2 000

0

A f g h a

n i s t a n

B a n g l a d

e s h

B h u t

a n I n

d i a

C a m b o

d i a

S r i L a

n k a

M a l d

i v e s

P a k i s

t a n

N e p a l

E X P Y

( 2

0 0 0

P P P $ )



37

100

Per cent

90

80

70

60

50

40

30

20

10

0PRODY < 2000

2 000 < = PRODY < 5 000

5 000 < = PRODY < 10 000

10 000 < = PRODY < 15 000

15 000 < = PRODY < 20 000

20 000 < = PRODY

Missing PRODY

A f g h a n

i s t a n

B a n g l a

d e s h

B h u

t a n

I n d i a

C a m b o

d i a

S r i L a

n k a

M a l d i

v e s

P a k i s

t a n

N e p a

l

Figure 18. Nepal in comparative perspective: Export sophistication 2010



Figure 17. Nepal’s export sophistication


2010

2009

2003

2000

1995

1992

1988

1985

1982

0 20 40 60 80 100

PRODY < 2 000

2 000 < = PRODY < 5 000

5 000 < = PRODY < 10 000

10 000 < = PRODY < 15 000

15 000 < = PRODY < 20 000

20 000 < = PRODY

Per cent



38

income and employment, and b) to upgrade the industrial and export structures and

capabilities in order to produce and export (or expand the production and export of) more

sophisticated products requiring a greater number of capabilities.

Nepal exported 65 products (SITC, Rev. 2, 4-digit levels) in 1995, exhibiting

comparative advantage in 42 of them (as measured by the revealed comparative

advantage (RCA) index). The number of items increased to 343 in 2003, with revealed

comparative advantage in 127 items. However, while the number of items exported

increased to 438 by 2010, the rate of increase is strikingly slower. Moreover, the number of

products with revealed comparative advantage declined slightly to 120.

The Open Forest represents products which Nepal did not export intensively, as

measured by the RCA index. Of the 733 products in the Open Forest in 1995, 193 products

were “nearby” the current exports of that time. Comparative advantage was gained in

a quarter of the nearby products by 2003, as well as 7 faraway products and 34 middle-distance products. Comparative advantage was thus gained in 12.4 per cent of the

unexploited Open Forest products. Concurrently, comparative advantage was lost in just

nine products, three fourths of which had low Path or PRODY (or both). The median

PRODY and Path of the 91 products in which comparative advantage was gained were,

respectively, 17 per cent and 29 per cent greater than that of the entire export basket of

1995. Along with the robust increase in export sophistication, the prospects for future export

diversification also increased substantially during 1995-2003. The value of the Open Forest

trebled to 1.2 million (PPP$ 2000). In contrast, during 2003-2010 comparative advantage

was gained in 42 products or 6.5 per cent of the unexploited Open Forest products in the2003 (compared to the gain made earlier the rate was 50 per cent slower). Out of them,

18 are nearby, 17 middle-distance and 7 faraway products. Thus, comparative advantage

was gained in less than 10 per cent of the nearby products. The median PRODY and Path

of the 42 products (in which comparative advantage was gained) is only 5.5 per cent and

0.6 per cent, respectively. This is greater than that of the entire export basket of 2003,

implying that the gain in comparative advantage during 2003-2010 entails a relatively less

gain in sophistication and connectedness of products, compared to the gain in comparative

advantage during 1995-2003. In the recent period (2003-2010), Nepal lost comparative

advantage in 49 products, with a concentration of medium-to-high connected products.Some two thirds of them25 had low PRODY or Path or both, compared to three fourths in

the earlier period.

25 PRODY and Path data are available for only 47 of the 49 products in which comparative advantage was

lost.



39

Table 6 shows the features of Nepal’s export basket in 2010. Of the 120 products, in

which the country had comparative advantage in 2010 and which accounted for 95 per cent

of total exports, 31 belonged to the capital-intensive category, 27 to the labour-intensive

category and 20 to the cereals category. There are only 6 and 8 products, respectively, in

the machinery and chemical categories. Nepal’s exports are grouped into 9 Leamer categories, excluding petroleum where no products are exported with comparative

advantage. The country’s exports have above-average PRODY and Path in 4 and 6

Leamer categories, respectively.26

While the value of Open Forest trebled during 1995-2003, it increased by just

3 per cent during 2003-2010. Of the 656 products in the Open Forest in 2010, about

28 per cent (182) are nearby products, about 40 per cent (251) are middle-distance

products and the remaining 32 per cent (222) are faraway products.27 Tables 7-9 show the

top 10 nearby, middle-distance and faraway products ordered by Strategic Value. As

expected of the Open Forest of a low-income country with a weak industrial base, Strategic

Value of Open Forest products vary inversely with density, shown in Figure 19.28 Density

measures the relative proximity between Open Forest products from the current export

basket. In other words, on average, the higher the Strategic Value of a product, the farther

it is from the current export basket.

The overall top 30 products, as well as the top 30 nearby products (not shown here,

but shown in Appendix Tables A1 and A2) have high Path. This means they are highly

interconnected with other products, and gaining comparative advantage in them would

augur well for export diversification and structural transformation. None of the overall top30 products in the Open Forest are nearby, mostly being far away from Nepal’s current

industrial and export capabilities. But even among the nearby products, there are products

that have medium-to-high sophistication as well as high connectedness with other

products, and represent relatively high Strategic Value. Of the 182 nearby products, there

are almost an equal number of capital and labour intensive products (32 and 31,

respectively). This is followed by tropical agriculture and cereal etc. (27 each), animal

products (24) and forest products (14) (table 10). There are only 8 and 4 products in the

machinery and chemical categories, respectively. These two are mostly faraway products.

Among the top 30 nearby products, 14 are capital intensive, 6 labour intensive, 4 forest,3 machinery, 2 tropical agriculture and 1 cereal products. Given the current industrial and

export capabilities, it is possibly easier to export the nearby products. Notably, nearby and

middle-distance products with relatively high path and Strategic Value includes textiles,

such as fabric and yarn (although capital intensive). These are consistent with certain types

of fabric and yarn already being exported with comparative advantage, implying capabilities

for such production. However, the presence of petroleum products as nearby products is

likely to be a spurious result, given Nepal’s natural resource endowment.

26 Classification of products under Leamer (1984) groups is based on the classification provided in the data

set of Hidalgo et al. (2007). Leamer group name is missing for one product.

27 Leamer group name is missing for one product.

28 The correlation coefficient is 0.49 and is significant at 1 per cent level.



40

Table6.Featuresofexportbasketin2010

Leamergroup

No.of

No.of

Exports,

No.of

Sharein

Average

Average

Average

Average

products

tota

l

2010

produc

ts

exports

PRODY

Path

PRODY

Path

exported

produ

cts

(’000$)

with

(All,

(All,

RCA>1

Leamer)

Leamer)

Petroleum

1

10

0.40

0

0.00

11,437.13

142.19

9,554.66

111.96

Rawmaterials

19

62

39,389.06

5

4.73

9,018.40

128.27

10,060.59

105.34

Forestproducts

25

38

13,750.14

8

1.65

13,559.48

148.99

13,134.52

135.58

Tropicalagriculture

33

46

125,323.20

8

15.05

7,884.47

125.69

7,553.05

115.01

Animalprodu

cts

31

53

32,078.28

7

3.85

10,326.37

123.78

10,813.78

119.38

Cereals,etc.

45

81

32,878.50

20

3.95

9,248.02

109.89

9,008.84

102.38

labourintens

ive

74

96

118,530.48

27

14.23

10,661.69

132.38

11,333.09

135.67

Capitalintensive

79

115

412,895.66

31

49.58

11,199.42

149.34

11,471.45

150.74

Machinery

93

179

21,251.32

6

2.55

16,141.82

140.72

16,054.66

136.10

Chemical

38

94

36,664.20

8

4.40

14,931.70

149.49

15,274.56

142.16

Total

438

774

832,761.21

120

S o u r c e :

Auth

or’scalculation.



41

Table7.Top10nearbyproducts,orderedbyStrategicValue

SITC

Product

Leamer

Path

PRODY

PRODY

Path

Contribution

Strategic

code

group

tercile

tercile

toOpen

Value

Forest(PRODY*

density)

8932

Sanitaryortoiletart.

labour

195.22

13,541.54

medium

high

2,289.36

15,076.17

ofmaterialsofdiv.58

intensive

6421

Boxes,bags&oth.packing

Forest

182.21

14,854.13

medium

high

2,576.47

14,184.04

containers,ofpaper/papbd

products

6924

Casks,drums,boxesof

Capital

183.55

12,609.02

medium

high

2,187.44

14,059.19

iron/steelforpackinggoods

intensive

6417

Paper&paperboard,

Forest

183.28

10,462.18

medium

high

1,814.01

13,886.84

corrugated,creped,

products

crinkledetc.

6794

Castingsorironorsteel,

Capital

185.55

11,749.49

medium

high

2,065.54

13,790.18

intheroughstate

intensive

8212

Furnitureformedical,surgical,

Labour

176.70

13,331.69

medium

high

2,296.04

13,489.34

dentaletc.practice

intensive

6975

Sanitarywareforindooruse,

Capital

173.04

15,926.95

high

high

2,716.29

13,319.42

andparts

intensive

6973

Domestic-type,non-electric

Capital

179.19

12,795.86

medium

high

2,324.09

13,104.04

heating,cookingappar.

intensive

6517

Yarnofregeneratedfibres,

Capital

178.03

9,639.35

low

high

1,804.43

13,042.27

notforretailsale

intensive

6560

Tulle,lace,embroidery,

Capital

171.36

12,836.79

medium

high

2,497.21

12,899.75

ribbons,&othersmallware

s

intensive

S o u r c e :

Auth

or’scalculation.



42

Table8.Top10middle-distancepr

oducts,orderedbyStrateg

icValue

SITC

Product

Leamer

Path

PRODY

PRODY

Path

Contribution

Strategic

code

group

tercile

tercile

toOpen

Value

Forest(PRODY*

density)

6996

Miscellaneousarticlesof

Capital

207.69

15,704.52

high

high

2,418.90

17,006.43

basemetal

intensive

6785

Tube&pipefittings(joints,

Capital

207.19

15,323.85

high

high

2,385.44

16,929.91

elbows)ofiron/steel

intensive

7449

Partsofthemachinery

Machinery

199.53

16,022.39

high

high

2,239.96

16,886.23

of744.2-

7139

Partsofint.comb.piston

Machinery

194.09

16,920.96

high

high

2,314.52

16,883.36

enginesof713.2-/3-/8-

8935

Art.ofelectriclightingof

Labour

198.18

15,538.38

high

high

2,341.37

16,768.43

materialsofdiv.58

intensive

6210

Materialsofrubber

Capital

198.81

15,040.43

high

high

2,219.33

16,665.21

(e.g.pastes,plates,sheets,etc.)

intensive

6953

Othertoolsforuseintheha

nd

Capital

189.71

14,733.62

medium

high

2,200.75

16,174.98

intensive

6991

Locksmithswares,safes,

Capital

189.13

16,414.08

high

high

2,416.89

16,155.87

strongroomsofbasemetal

intensive

7919

Rail&tramwaytrackfixture

s&

Machinery

191.91

14,949.81

high

high

2,077.53

16,139.35

fittings,signall.equi.

6424

Paperandpaperboard,

Forest

190.87

15,599.70

high

high

2,222.51

16,131.59

cuttosizeorshape,n.e.s.

products

S o u r c e :

Auth

or’scalculation.



43

Tab

le9.Top10farawayprodu

cts,orderedbyStrategicV

alue

SITC

Product

Leamer

Path

PRODY

PRODY

Path

Contribution

Strategic

code

group

tercile

tercile

toOpen

Value

Forest(PRODY*

density)

5335

Colour.preptnsofakindus

ed

Chemical

196.55

18,099.09

high

high

2,307.83

17,717.46

inceramic,enamelli.

6632

Naturalorartificialabrasive

Labour

194.49

18,839.34

high

high

2,423.08

17,615.70

powderorgrain

intensive

7267

Otherprintingmach.

Machinery

174.30

21,054.27

high

high

2,200.40

17,527.86

forusesancillarytoprinting

7492

Taps,cocks,valvesetc.

Machinery

189.17

17,626.42

high

high

2,209.09

17,306.65

forpipes,tanks,vatsetc.

7269

Partsofthemachines

Machinery

176.80

20,564.04

high

high

2,337.46

17,207.22

of726.31,726.4-,726.7-

6418

Paper&paperboard,

Forest

189.02

19,418.90

high

high

2,430.43

17,197.90

impregnat.coat.

products

surface-coloured

7849

Otherparts&accessories

Machinery

193.79

17,158.54

high

high

2,166.88

17,185.51

ofmotorvehicles

7429

Partsofthepumps&liq.

Machinery

176.21

18,779.22

high

high

2,127.31

17,088.95

elevatorsof742–

7439

Partsofthemachines

Machinery

181.99

16,500.97

high

high

1,975.90

17,027.87

of743.5-,743.6-

7493

Transmissionshafts,cranks

,

Machinery

171.15

17,263.96

high

high

1,855.28

16,647.64

bearinghousingsetc.

S o u r c e :

Auth

or’scalculation.



44

Figure 19. Strategic Value against density of unexploited products, 2010


Table 10. Products in Open Forest according to Leamer classification

All products Nearby products

in Open Forest

Petroleum 10 5

Raw materials 57 12

Forest products 30 12

Tropical agriculture 38 27 Animal products 46 24

Cereals, etc. 61 27

Labour intensive 69 32

Capital intensive 85 31

Machinery 173 8

Chemical 86 4

Total 655 182


20 000

15 000

10 000

5 000

0

.05 .1 .15 .2 .25 .3

Density

S t r a t e g i c_ v a l u e



45

G. Analysis of targeted products

Two groups of targeted products are comparatively analysed: a) NTIS-identified

products, which include some products identified in the Trade Policy, b) products exclusive

to Trade Policy (not in NTIS). The approach is to first assess export implications of diversification and structural transformation in promoting the products identified by

NTIS – which is the trade strategy of the Government of Nepal until at least 2015. The

NTIS-identified products are then compared with products identified by Trade Policy.

Researchers compare the identified products with the existing export basket, as well as the

Open Forest. The “other” export potential products of NTIS are also briefly assessed.

1. NTIS products

There are 131 NTIS-identified products in terms of HS (2002) 6-digit codes and 61

in terms of SITC (Rev. 2) 4-digit codes. Although, less product diversity is captured by SITC

codes than HS codes, the former is still employed as the PRODY and Path data available

are based on SITC classification. Almost all of iron and steel products are manufactured,

capital-intensive, numbering at 38 of the 61 identified, followed by wool products with 11.

Accorded special priority in NTIS, agro-food products number 7.29 Forty-seven of the 61

identified products are exported in 2010, accounting for 37 per cent of Nepal’s merchandise

exports. All the 14 products identified, though not exported in 2010, are iron and steel

products. Twenty-four products are exported with comparative advantage in 2010. Notably,

natural honey, silver jewelry and handmade paper are not exported with comparative

advantage. Twenty-two of the 61 products are also listed in the Trade Policy 2009. Instant

noodle and all iron and steel products are exclusive to NTIS. Table 11 summarizes the

features of NTIS-identified products.

Of the 61 identified products, 36 have PRODY greater than EXPY (measuring

overall sophistication of the export basket) for 2010. Twenty-four products have PRODY

and Path greater than the average for their respective Leamer groups. However, these are

disproportionately concentrated in the iron and steel products category.

As expected, basic agricultural products, such as lentils, cardamom, ginger, tea,medicinal herbs and honey, have low PRODY. This points to it being less than EXPY, and/

or belonging to the lowest PRODY tercile (table 11 and figure 20). Notably, while PRODY

represents the sophistication of a product by taking into account the income levels of

countries effectively exporting it, Path shows how connected a product is with other

products. A higher Path indicates greater potential for industrial and export diversification.

The highest Paths are depicted by, on average, handmade paper, iron and steel products,

natural honey and pashmina products (figure 21). Outside of iron and steel products, only

handmade paper makes it to the top (first) Path tercile. Among agro-food products, natural

29 Essential oils are classified as chemical products under Leamer classification, but NTIS considers them

agriculture products. They are derived from plants and the WTO’s Agreement on Agriculture lists them as

agriculture goods. Another incongruity is that medicinal plants are placed under animal products by Leamer.

But these do not affect the analysis in any significant way.



46

T

able11.Summaryfeatures

ofNTIS-identifiedproducts

NTIS

Leamergroup

No.of

No.of

Products

Exports

Sharein

PRO

DY>

RCA>1

Path

PRODY

group

identified

products

exported

2010

NTIS

EX

PY

tercile

andpath>

products

alsoin

($’000)

exports

Average

TradePolicy

Leamer

Lentils

Tropicalagriculture

1

1

1

51,394.18

16.72

0

1

low

0

Cardamom/

Tropicalagriculture

1

1

1

22,929.11

7.46

0

1

low

0

ginger

Tea

Tropicalagriculture

1

1

1

16,356.39

5.32

0

1

low

0

Medicinalherb

s

Animalproducts

1

1

1

6,020.03

1.96

0

1

low

0

Instantnoodle

Cereals,etc.

1

0

1

5,967.00

1.94

1

1

low

0

Essentialoils

Chemical

1

1

1

504.29

0.16

0

1

medium

0

Naturalhoney

Tropicalagriculture

1

1

1

3.14

0.00

1

0

medium

1

Ironandsteel

Capital

38

0

24

146,624.80

47.70

3

1

7

High(27)

,

22

products

intensive(37),

medium(10)

,

rawmaterials(1)

low(1)

Woolproducts

Labourintensive

11

11

11

34,233.23

11.14

1

9

Medium(6)

,

0

low(5)

Pashmina

Capital

2

2

2

19,882.70

6.47

0

2

Medium

0

products

intensive(1),

labour

intensive(1)

Silverjewelry

Labourintensive

2

2

2

2,986.02

0.97

1

0

Low

0

Handmade

Forestproducts

1

1

1

513.00

0.17

1

0

High

1

paper

Total

61

22

47

307,413.89

100.00

3

6

24

24

S o u r c e :

Auth

or’scalculation.



47

Figure 20. Sophistication and connectivity of NTIS-identified products


Note: The dashed lines demarcate high, medium and low PRODY/Path groups.

honey and essential oils stand out for belonging to the medium (second) Path tercile, the

rest have low Path. This is despite all identified agro-food products having low PRODY.

Appendix Table A3 provides detailed data on NTIS-identified products.

The NTIS-identified products are compared with the export basket of 2010, which

comprises of all products that are exported regardless of comparative advantage

(table 12). In terms of sophistication (PRODY) and connectedness with other products

(Path), this reveals that, on average, the identified products fare better than the export

basket (all exported products or products exported with comparative advantage). However,

this is overwhelmingly driven by the presence of iron and steel products among the

identified products. Excluding iron and steel products, the identified products, on average,

mostly lag behind the export basket on the same parameters. Apart from most iron and

steel products, handmade paper, pashmina products and three wool products have Path

greater than the export basket average.

Of the 61 identified products, 37 are not exported with comparative advantage,

meaning that they are strictly in the Open Forest. The median Path and Strategic Value of

the 37 products (31 of which are iron and steel products) are higher than those of the entireOpen Forest. This is driven by the presence of a high number of iron and steel products

(table 13). Handmade paper also has Path and Strategic Value greater than the median of

the Open Forest, and greater than the majority of iron and steel products. Natural honey,

20 000

15 000

10 000

5 000

0

50 100 150 200Path

P R O D Y



48

Table 12. Summary statistics of PRODY, Path and Strategic Value of across groups

Variable No. of Mean Median Standard Min Max

products Deviation

Export basket, 2010

PRODY 434 12,008.96 12,351.52 5,206.23 1,371.40 34,113.69

Path 434 136.04 137.50 34.13 9.28 207.69

RCA>1

PRODY 119 9,549.24 9,148.28 4,659.04 1,371.40 19,962.62

Path 119 131.57 132.83 33.33 30.19 197.11

NTIS products

PRODY 61 10,424.90 10,939.27 3,894.29 1,371.40 20,158.88

Path 61 146.84 144.30 31.44 86.31 207.19

Strategic Value 61 11,094.15 11,284.76 3,227.76 5,889.44 16,929.91

NTIS iron and steel

PRODY 38 12,116.09 12,636.83 2,765.47 6,290.71 17,100.74

Path 38 163.28 168.31 25.14 105.91 207.19

Strategic Value 38 12,999.33 13,545.04 2,238.39 8,271.48 16,929.91

NTIS excluding iron

and steel

PRODY 23 7,630.75 7,167.00 3,923.76 1,371.40 20,158.88

Path 23 119.68 115.95 19.88 86.31 171.62

Strategic Value 23 7,946.47 7,852.64 1,832.93 5,889.44 14,714.97

Open Forest

PRODY 656 12,889.38 13,391.83 5,445.34 801.23 34,113.69

Path 656 129.97 133.49 38.31 2.89 207.69

Strategic Value 656 11,074.84 11,517.18 3,474.22 180.99 17,717.46

NTIS products in

Open Forest

PRODY 37 11,828.01 12,322.83 3,403.55 6,290.71 20,158.88

Path 37 156.13 164.60 29.08 98.70 207.19Strategic Value 37 12,327.22 13,280.49 2,869.61 6,488.86 16,929.91


Note: There are 434 products in the 2010 export basket in this calculation, instead of 438. This isbecause PRODY and Path values are not available for 4 of the exported products.

silver jewellery (2) and wool products (2) have Strategic Value lower than the median.

However, among them, natural honey has relatively higher Strategic Value along with Path.

Of the 37 products, 21 are middle-distance products, 12 are nearby products and 4 are

faraway products – all four are iron and steel products (table 13). Seven of the 12 nearby

products have medium-to-high Path and Strategic Value, which is higher than the medianof nearby products in the Open Forest.



49

The Strategic Values of all 61 identified products are considered in order to gauge

what opportunities they have to offer, regardless of whether they were exported with

comparative advantage or not. This is in terms of producing and exporting other more

sophisticated products once the country produces and exports the identified products

effectively. “Effectively”, in this context, means production and export capacity in the

identified products is significantly scaled up. It is unnecessary to determine a calculation

threshold for effective producing and exporting of the identified products, as all identified

products are being considered. It is assumed that production and export capacity in these

are less than the desired level, despite the potential. This is why they are being targeted for

export promotion. As a basis to determine what the “other” products are, the identified

products are defined as those in which the country does not have revealed comparative

advantage, i.e. RCA<1. Notably, the exercise findings reveal that the Strategic Value of

about half of the identified products is less than the average Strategic Value of Open Forest

products. All those with Strategic Value above average are iron and steel products, with the

exception of handmade paper (which has higher Strategic Value than most iron and steel

products).

Focusing on Path, we find that, for natural honey, pashmina products (1) and wool

products (3), Path is greater than Open Forest average. Only handmade paper and most of

the iron and steel products both have Strategic Value and Path greater than the average for

the Open Forest. Comparing the identified products with nearby products in the Open

Forest, we find that Path is greater than the Open Forest average also for essential oils,

handmade paper, natural honey, two pashmina products and six wool products. This is

besides most iron and steel products. Strategic Value is also greater than the Open Forest

average for essential oils, handmade paper, natural honey, one pashmina product, and one

silver jewellery product. Lastly, both Path and Strategic Value are also greater than the

Open Forest average for handmade paper and one pashmina product (table 14).

Table 13. NTIS products that lie in the Open Forest

NTIS group Faraway Middle Nearby Total Nearby products with

medium-to-high Path

and Strategic

Value>Median_nearbyOpen Forest

Handmade paper 0 1 0 1

Iron and steel 4 20 7 31 6

products

Natural honey 0 0 1 1 1

Silver jewellery 0 0 2 2

Wool products 0 0 2 2

Total 4 21 12 37 7




50

2. Identified in the 2009 Trade Policy

Products exclusive to the Trade Policy 2009 are analysed here. Trade Policy 2009

identifies products broadly, and not in terms of a standard classification system. Therefore,338 products are identified at HS (2002) 6-digit level, corresponding to the broad categories

identified in Trade Policy. This is based on HS 6-digit products exported in 2010 and HS

6-digit products under the same or similar categories considered in the International Trade

Centre’s Nepal Export Potential Assessment (ITC, 2007).

When converted to SITC (Rev. 2) 4-digit levels, the number reduces to 54. The

PRODY, Path and Strategic Value of these 54 Trade Policy-identified products are, on

average, less than the NTIS products. They are higher if compared to the NTIS product set,

excluding iron and steel products (table 15).30

The PRODY and Path of the Trade Policyproducts are, on average, also less tan those of the 2010 export basket.

Table 14. Comparison of NTIS products with export basket and Open Forest

NTIS group Path> Path> SV>Median_ Path &

Median_ Median_ Open Forest SV>Median_

export basket Open Forest Open Forest

Handmade paper 1 1 1 1

Iron and steel products 30 32 28 28

Natural honey 1

Pashmina products 1 1

Wool products 3 3

Total 35 38 29 29

NTIS group Path>Median_ SV>Median_ Path & SV>Median_nearby

nearby Open nearby Open Open Forest

Forest Forest

Essential oils 1 1

Handmade paper 1 1 1

Iron and steel products 37 34 30

Natural honey 1 1

Pashmina products 2 1 1

Wool products 6

Silver jewellery 1

Total 48 39 32


30 Appendix Table A4 provides details of the TP products.



51

In 2010, 48 of the 54 products are exported, 18 of those with comparative

advantage (table 16). The exports amounted to 12 per cent of total merchandise exports.

For 2010, 23 products have PRODY greater than the EXPY. Thirteen products have

PRODY and Path greater than the average for their respective Leamer groups. Incomparison to NTIS products, the Trade Policy products represent a greater variety of

manufactured goods.

Table 15. Summary statistics of Trade Policy identified products

Variable No. of Mean Median Standard Min Max

products Deviation

PRODY 54 8,902.05 8,232.19 4,742.03 1,576.46 20,806.53

Path 54 125.35 118.59 34.42 63.90 190.87

Strategic Value 54 9,198.65 8,471.46 3,364.40 3,792.50 17,197.90


Table 16. Summary features of Trade Policy identified products

Trade Policy Leamer No. of Products RCA>1 Exports PRODY> PRODY

group group products exported (U$ ’000) EXPY and Path>

Average_

Leamer

Coffee Tropical 1 1 0 316.14 0 0

agriculture

Floriculture Animal 2 2 0 285.64 1 0

products

Fresh Tropical 3 3 0 202.09 2 2

vegetables agriculture

Gems and Labour 2 1 0 0.63 1 0

jewellery intensive

Handicraft Capital 2 2 2 1,010.40 1 0

(non-wood) intensive (1),labour

intensive (1)

Handicraft Forest 1 1 1 3,293.58 0 0

(wood) products

Leather goods Capital 6 4 0 468.89 3 1

intensive (3),

labour

intensive (3)

Orange Tropical 1 1 0 3.21 0 0

agriculture

Paper products Forest 10 9 2 4,612.27 10 7

products



52

Of the 54 Trade Policy products, 12 have high Path and 17 medium Path (table 17).Three of the eight agricultural products have medium Path. All 11 RMG products have low

PRODY, but five of them have medium Path. A total of 18 products – including leather

goods, paper products, processed leather, vegetable seeds, towel and non-wood handicraft

– have medium-to-high PRODY and Path (figure 22).

There are 16 products with Path above the median Path of the 2010 export basket.

There are 13 products with Path and/or Strategic Value exceeding the median of NTIS

products. However, 29 products have Path and/or Strategic Value exceeding the median of

the NTIS product set, excluding iron and steel (table 18). While 21 products have Path

greater than the Open Forest median, 12 have Strategic Value greater than the Open

Forest median. The latter number increases to 27 or half of the Trade Policy products, if

comparison is made with the median of nearby products in the Open Forest. In particular,

leather goods, processed leather, paper products and towels exhibit higher-than-average

Path and Strategic Value consistently across all comparator groups. In contrast, products

such as coffee, floriculture, wooden handicraft, orange, silk products and woolen carpet

have below-average Path and Strategic Value across all comparator groups. Most RMG

products have lower-than-average Path and Strategic Value across most comparator

groups. However, compared to the NTIS product set (excluding iron and steel), five RMG

products exhibit higher-than-average Path and two higher-than-average Strategic Value.

This means that although they are low-tech labour-intensive manufactured goods, RMG

products hold greater potential for industrial and export diversification than many of the

NTIS products, excluding iron and steel.

Processed Capital 5 3 2 8,041.43 1 0

leather intensive

RMG Labour 11 11 7 21,336.80 0 0

intensive

Silk products Capital 3 3 1 9.09 0 0

intensive (1),

cereals,

etc.* (1)

Towel Capital 5 5 2 2,058.14 3 2

intensive

Vegetable Animal 1 1 0 37.38 1 1

seeds products

Woolen Capital 1 1 1 59,675.00 0 0

carpet intensive

Total 54 48 18 101,350.68 23 13


Table 16. (conitued)

Trade Policy Leamer No. of Products RCA>1 Exports PRODY> PRODY

group group products exported (U$ ’000) EXPY and Path>

Average_

Leamer



53

Figure 21. Sophistication and connectivity of Trade Policy products


Note: The dashed lines demarcate high, medium and low PRODY/Path groups.

Table 17. Distribution of Trade Policy identified products across path terciles

Trade Policy group High Medium Low Total

Coffee 0 0 1 1

Floriculture 0 0 2 2Fresh vegetables 0 2 1 3

Gems and jewellery 0 0 2 2

Handicraft (non-wood) 0 1 1 2

Handicraft (wood) 0 0 1 1

Leather goods 2 2 2 6

Orange 0 0 1 1

Paper products 8 1 1 10

Processed leather 0 3 2 5

RMG 0 5 6 11

Silk products 0 0 3 3

Towel 2 2 1 5

Vegetable seeds 0 1 0 1

Woolen carpet 0 0 1 1

Total 12 17 25 54


20 000

15 000

10 000

5 000

0

P R O D Y_ T

P

50 100 150 200

Path_TP



54

Notably, clothing has been the launching pad of industrialization and export

diversification in many developing countries. Even among agricultural products, freshvegetables and vegetable seeds have above-average Path and Strategic Value when

compared to the NTIS product set, excluding iron and steel. After pressure from the

Garment Association of Nepal, in September 2012 the government decided to include RMG

in the NTIS list of priority products. However, strategies specific to that sector are yet to be

formulated.31

Finally, in combining NTIS products and Trade Policy products (excluding iron and

steel, which yields 77 products), findings show that among the top 20 products in terms of

Strategic Value, 19 products are exclusive to Trade Policy, while one product is common toboth. In sum, quite a few of the products identified by Trade Policy hold greater potential for

export diversification and structural transformation than many NTIS products, particularly

when excluding iron and steel.

In 2010, 36 of the Trade Policy products were not exported with comparative

advantage, hence strictly belong to the Open Forest (table 19). Three fourths of them are

nearby products, including manufactured products such as leather goods, paper products

and towel. These have relatively high sophistication, connectedness and/or Strategic Value.

About a dozen of the Trade Policy-identified nearby products have medium-to-high Path

and Strategic Value. This is greater than the average of nearby products in the Open Forest

Table 18. Number of products exceeding median values of comparator groups

Trade Policy group Path> Path> SV> Path> SV> Path> SV> SV>

Export NTIS NTIS NTIS- NTIS- Open open Nearby_

basket iron, iron, Forest forest open

steel steel forest

Fresh vegetables 0 0 0 2 2 2 0 2

Gems and jewellery 0 0 0 0 1 0 0 1

Handicraft (non-wood) 0 0 0 1 2 1 0 1

Leather goods 2 2 2 4 4 3 2 4

Paper products 9 8 8 9 10 9 8 10

Processed leather 1 0 1 3 3 1 0 3

RMG 1 0 0 5 2 2 0 1

Towel 3 3 2 4 4 3 2 4

Vegetable seeds 0 0 0 1 1 0 0 1

Total 16 13 13 29 29 21 12 27


31 “Govt to include garment in NTIS,” Republica, 8 September 2012, http://www.myrepublica.com/portal/

index.php?action=news_details&news_id=41317



56

Table20.Featuresof

“other”NTISproducts

SITC

Product

Leamer

PRODY

Path

Strategic

P

RODY

Path

Distancefrom

code

group

value

tercile

tercile

current

exp

ortbasket

223

Milk&cream,fresh,

Animal

15,617.02

156.01

12,934.60

high

high

middle

no

tconcentratedorsweetened

products

224

Milk&cream,preserved,

Animal

15,704.30

131.91

10,443.87

high

medium

middle

co

ncentratedorsweetened

products

230

Butter

Animal

18,586.19

143.02

12,161.97

high

medium

products

240

Cheeseandcurd

Animal

18,067.83

167.50

13,253.28

high

high

middle

products

611

Sugars,beetandcane,

Tropical

6,461.67

74.16

4,554.49

low

low

nearby

ra

w,solid

agriculture

612

Refinedsugarsand

Tropical

5,718.05

138.51

9,740.18

low

medium

nearby

otherprod.ofref.beet/cane

agriculture

6612

Portlandcement,ciment

Labour

9,491.74

140.98

9,391.88

low

medium

Error!

Bookmark

fondu,slagcementetc.

intensive

notde

fined.

7711

Tr

ansformers,electrical

Machinery

12,421.36

151.66

10,603.38

medium

high

S o u r c e :

Auth

or’scalculation.



57

products. Interestingly, whereas agricultural products generally depict lower-than-average

PRODY, Path and Strategic Value, the four dairy products surpass the majority of the main

NTIS products on these attributes. Note that powder milk33 production is a priority industry

as per Industrial Policy. This can also be said of cement production, which despite having

low PRODY, has medium Path, greater than the average main NTIS product. Except rawsugar, all “other” NTIS products have medium-to-high Path. Two (butter and transformer) of

the 8 products were exported with comparative advantage in 2010. The remaining six

belong to the Open Forest – 3 middle-distance from and 3 nearby the current export basket

(containing goods exported with comparative advantage).

H. Discussion

With regard to its export sector, the challenges facing Nepal can be broadly

grouped under two clusters. The first challenge is to increase the quantum (volume andvalue) of exports so as to stop the burgeoning trade deficit. More importantly, to create

employment and generate income for an expanding labour force, raise economic growth

and alleviate poverty. This first challenge is more pressing and has to be addressed

immediately. The second challenge is to diversify and increase the sophistication and

complexity of the export basket, that is, to produce and export a wider range of new and

increasingly sophisticated and complex products.

Addressing the second challenge takes a longer period of time than the first, as it

pertains to structural transformation of an economy. Also, it may not immediately addressthe first challenge. Given resource constraints (among other things), at times policymakers

could face a trade-off between according priority to tackling the two challenges. However,

addressing the second challenge is critical for ensuring sustained per capita income growth

over the long run, as mounting evidence shows. This is discussed in Section 3.

Policymakers should explore ways to increase the sophistication of exports when

strategizing and directing resources to develop the sector. This would enable meeting the

urgent objectives of immediate income and employment generation, as well as poverty

alleviation.

Considerations of export potential and socio-economic impact underlie the choice of

products for priority development and promotion in policy, notably Trade Policy 2009 and

NTIS 2010. Export potential points to mainly demand, market access and competitiveness

conditions, and supply-side capacities, while socio-economic impact is mainly employment

and income generation, poverty reduction, and backward linkages.

Attention to socio-economic impact is guided by national development goals and

objectives. Attention to export potential reflects realism. Products to be targeted for export

promotion must have favourable demand and market access conditions. They must be

fairly competitive, and there must exist (or there must be a strong potential for the

emergence of) a domestic supply capacity associated with these products. In view of the

33 It is not captured individually under SITC 4-digit classification.



58

power crisis (which could act as a critical supply-side constraint), NTIS takes into account

the electricity intensity in production. All, but iron and steel products, have low-to-medium

electricity intensity.

While Trade Policy 2009 provides the overall trade-related policy direction, it is

NTIS 2010 that is to guide the Government’s trade-related priorities until 2015. The

Government has allocated budget specifically for NTIS implementation. Hence, the

products identified in NTIS (some of which are also in Trade Policy) would take precedence

over other Trade Policy products, in terms of priority accorded for development and

promotion. NTIS products differ in their sophistication and potential for structural

transformation. However, overall they are below-average on these parameters.

Iron and steel products have mostly high-to-medium sophistication and

connectedness, as well as above-average Strategic Value. Also, most of it lies in the “core”

of the product space. Agriculture products, such as tea, lentil, ginger, cardamom andmedicinal herbs are at the low end. Handmade paper, some pashmina and wool products,

honey and essential oils offer better prospects. Trade Policy products (products exclusive

to Trade Policy) are also overall below-average, but fare better than some NTIS products.

This is particularly when iron and steel products are excluded from the NTIS product set.

Leather goods, paper products, processed leather, vegetable seeds, towel and non-wood

handicraft hold prospects that are above average. These also surpass quite a few NTIS

products. Note that paper production has been listed as a priority industry in Industrial

Policy 2010. RMG and woolen carpets, still among the major exportable goods of the

country, are notably excluded from NTIS, although, they are on the Trade Policy list. Whilewoolen carpets (here captured by a single SITC code) are well below average, in terms of

sophistication and potential for structural transformation, a few RMG product categories

fare better than the average NTIS product. Thus, there is some scope to better the

prospects of achieving greater export sophistication and diversification by drawing in some

of the products identified by Trade Policy, but excluded from NTIS. Also to be noted is that

Trade Policy, while representing a greater range of manufactured products than NTIS, has

fewer capital-intensive products. Also, some of the “capital-intensive” products in Trade

Policy are, in practice, relatively labour-intensive in the Nepali context (e.g. carpet, silk

products, leather and leather products).

Of course, the choice of products for targeting is limited by the concentration of

Nepal’s effective exportable products (here defined as those with RCA>1) in the periphery

of the product space. In turn, this is responsible for nearby products in the option set of

unexploited opportunities (the Open Forest) being relatively low in sophistication and

Strategic Value. And nearby products are those that can be relatively easy to produce and

export effectively, given the economy’s current capabilities. The question is whether the

possibilities (however limited) for export diversification and structural transformation offered

by nearby products in the Open Forest are being optimally captured while targetingproducts.

Quite a few of the targeted products (39 per cent of NTIS products and 33 per cent

of Trade Policy products) are exported with comparative advantage, and hence do not



59

belong to the Open Forest in a strict sense. A dogmatic argument would be that such

products should not be targeted at all, as they do not lie in the Open Forest. But in practice,

it may not be sensible when targeting products to only include those not exported with

comparative advantage, thereby excluding all products exported with comparative

advantage. This is even if the basis for targeting is purely the potential for structuraltransformation (which is rarely the case). As the rationale for targeting, a product may be

exported with comparative advantage in a technical sense (defined as RCA>1). However,

there may be considerable need and scope to enhance the capabilities associated with its

production and exportation. This is especially with regard to increasing production and

exports.

Our approach, therefore, has been to regard all identified products as not being

effectively exported and then assess their sophistication, connectedness and Strategic

Value. The majority of NTIS and Trade Policy products have Strategic Value less than the

median of Open Forest products, while the opposite is true when the comparison is made

with only the median of Open Forest nearby products. However, in the latter case the

above-average products are mostly iron and steel. Furthermore, among the 39 NTIS and

Trade Policy products that lie strictly in the nearby area of the Open Forest, about half have

medium-to-high Path and Strategic Value greater than the median of all the nearby Open

Forest products.

However, it is notable that five NTIS products (out of 12 that are in the nearby area

of Open Forest) are among the top 25 nearby Open Forest products in terms of Strategic

Value. These are also only iron and steel products, and only four Trade Policy products (outof 27). Similarly, only 4 NTIS products are among the top 25 middle-distance Open Forest

products in terms of Strategic Value. Again, all iron and steel (out of 21 that are in the

middle-distance area of the Open Forest), and 2 Trade Policy products (out of 5). Likewise,

none of the NTIS products (out of four that are in the faraway area of the Open Forest) and

only 1 Trade Policy product (out of 4) is among the top 25 faraway Open Forest products in

terms of Strategic Value. Overall, out of the identified 73 products that lie in the Open

Forest (37 NTIS and 36 Trade Policy), only 2 products (one Trade Policy and one NTIS) are

among the top 25 products in the Open Forest in terms of Strategic Value.

Therefore, there are some unexploited products with relatively high potential for

further export diversification that should be considered while targeting products. This is, of

course, subject to the condition that certain minimum demand, supply-side and

competitiveness conditions are met. Attention must be paid to the fact that some nearby to

middle-distance products in the Open Forest with relatively high Strategic Value and

interconnectedness are textiles. Most are not featured on either NTIS 2010 or Trade Policy

2010. That these are close to Nepal’s current industrial and export capabilities is consistent

with the fact that exports of non-carpet textiles, like yarn and fabrics, have nearly trebled in

value during 2003-2011. This nearly offsets the sharp decrease in clothing exports, andmakes up about a quarter of merchandise exports in 2011. Even if it involves capital-

intensive production (as per Leamer classification) and depends largely on imported raw

materials, its capital-intensiveness and import dependence is unlikely to be higher than that

of iron and steel. These were targeted by NTIS 2010. Additionally, some of the “other”



60

export potential products identified by NTIS have greater connectedness and Strategic

Value than most of the “main” NTIS products, excluding iron and steel. These “other” export

potential products, such as dairy products and cement, are not accorded priority.

The challenge in targeting lies in striking a balance between achieving short-term

efficiency and realization of urgent socio-economic goals, and the long-term imperative of

structural transformation. While these short-term goals aim to follow the signals of

comparative advantage, reliance on comparative advantage alone will not generate

structural transformation.

Given current capabilities, it is neither feasible nor desirable to attempt to “jump” to

the so-called “core” of the product space (mainly machinery and chemicals). Nepal’s

position at the “core” is sparse, except for iron and steel. That would be taking the

comparative advantage-defying industrialization strategy to the extreme (see Lin and

Chang 2009 for a debate on industrialization strategy). The success of such a strategy isfraught with high uncertainty, even as scarce resources are diverted away from sectors –

i) that exhibit comparative advantage or in which the acquisition of comparative advantage

will not exact high short-term costs; ii) that have high potential to realize urgent socio-

economic goals of, for example, employment generation and poverty reduction; iii) that

offer moderate prospects for further industrialization and export sophistication and

diversification, albeit not on the grand scale presented by “core” products.

Moreover, in the presence of production fragmentation and rising intermediate good

trade, outward sophistication or complexity of a final product does not tell the actual extent(or nature) of value addition and manufacturing that takes place in a particular country.

Products falling under SITC (Rev. 2) 2-digit codes 67 (iron and steel) and 69 (manufactures

of metals, nes) are among the so-called “core” products. The former accounted for about

15 per cent of Nepal’s merchandise exports in 2010, and all are targeted in NTIS. These

products augur well for structural transformation, having well-above-average PRODY, Path

and Strategic Value and leading the NTIS products on these parameters. But the industry is

totally dependent on imported raw materials, with little prospects for backward linkages,

and value added is roughly 20 per cent of total cost (GoN, 2010c).

Major exports include cold rolled steel sheets, galvanized color-coated steel sheets,

black galvanized steel pipes and black galvanized wires. But production is based on

imports of hot-rolled coil, sheets, steel wires, MS billets, sponge iron, and zinc (GoN,

2010c). Imports are mostly from India, and exports are destined mostly for India. To keep

abreast of international development, many companies invest in the latest technology in

most of the product lines on a regular basis. The companies also increase productivity and

quality and kept operating costs relatively competitive (GoN, 2010c). However, most of the

products are low-technology products, despite falling into the “core” of the product space.

The import dependence should not be held too much against the industry, particularly if

export volume is huge and growing, and given the relatively high sophistication and

connectedness of the products. But neither should the import dependence aspect be

dismissed when assessing the relative merits of targeting products. With scant domestic

raw material base, enhancing labour productivity and continuous technological up-



61

gradation for vertical product diversification and quality upgrading34 will be vital to making

exports of this sector sustainable.

The heavy industry should gear its development towards the domestic market to

create a foundation for breaking into export markets, if possible at all. Nepal’s iron and steel

industry, for one, predominantly caters to the domestic market and its first units were

established in the 1980s. Exports of iron and steel products have been notable only in the

last decade or so, and they are concentrated in the Indian market. Without the experience

of catering to the domestic market, breaking into the Indian market is likely to be difficult. It

follows that if effective exports of machinery (another “core” product) are to be aimed at,

then first the machinery industry for the domestic market should be developed, for example

agriculture tools and equipment, and industrial machinery. The machinery industry is listed

as a priority industry by the Industrial Policy 2010.

The method of product space analysis to assess structural transformationpossibilities is based on exports. It fails to capture the possibilities associated with import-

substituting sectors. For example, Nepal’s pharmaceutical industry caters significantly to

the domestic market in certain generic drugs (albeit in limited segments, mostly therapeutic

drugs of oral dosage forms – production of formulations from imported pharmaceutical

starting material).35 However, it has not been able to significantly break into export markets,

although industrialists say export potential exists.36 Preparations are underway by several

producers to export some pharmaceutical products.37 Note that pharmaceutical products

have quite high sophistication, connectivity with other products and Strategic Value, with

relatively high potential contribution to structural transformation. The product spaceanalysis does not recognize the country as possessing capabilities in pharmaceutical

production, as production is almost exclusively for the domestic market. For this reason,

they appear middle-distance or faraway from the current capabilities in the Open Forest.

This is neither an argument for or against targeting pharmaceuticals (although the Industrial

Policy lists them as a priority industry, and given Nepal’s endowment of medicinal plants

and heritage of traditional medicine, Ayurveda medicines may be a niche area in which

Nepal can develop competitive advantage). The point is that a comprehensive view of

structural transformation must also take into account the import-competing sectors. These

are important in their own right as tradable sectors and also for holding some possibilitiesfor exports.

As noted in the limitations section earlier, the methods employed in this paper do

not capture within-product sophistication, or quality of a product. While greater horizontal

34 This cannot be adequately captured by analysis at the SITC (Rev. 2) 4-digit level.

35 See Ministry of Health and Population/Government of Nepal and World Health Organization, Nepal

pharmaceutical country profile, September 2011; Budhathoki, Sushila, “Drug drive: The rising business of

pharmaceutical companies in Nepal”, Cover story, New Business Age, August 2012.

36 Umesh Lal Shrestha, President, Nepal Association of Pharmaceutical Producers and Managing Director,

Quest Pharmaceuticals Pvt Ltd, in an interview to Karobar national business daily (“Nepali pharmaceuticals are

of good quality yet low-priced”, 8 February 2011, p.12).

37 Oli, Sujan. “Four firms preparing to export pharmaceuticals within six months”, Arthik Abhiyan,

18 September 2012, p.1.



62

diversification and across-product sophistication are important, vertical diversification and

quality upgrading of existing exports are equally important. Within apparel, for instance,

“products exported by rich countries (or processes undertaken by them) – are likely to be

more skill and technology intensive, and yield higher wages and margins, than

standardized products exported by poor countries” (Lall et al., 2005). The case for verticalproduct diversification and quality upgrading is also critical for targeted products. NTIS

gives top priority to agro-food products, which are mostly unsophisticated products with

little connectedness with other products (see product space analysis above). However,

there is a window of opportunity for leveraging agro-food production for industrialization

purpose in the NTIS product-specific action matrix. It recognizes the need to achieve

horizontal and vertical product diversification, value addition and processing with regard to

the identified agro-food commodities. For example, in the case of cardamom, it strategizes

exploring avenues for product diversification, catering to spice, essential oil, cardamom

paper, incense, and color extraction industries. It saw high potential for diversification intoginger-based products (jam, jelly, candy, sauce, oleoresin). At present, even simple drying

of ginger is not generally done in Nepal. In the case of herbs and essential oils, NTIS

identifies the need to initiate R&D efforts towards processed products, such as perfumes,

food flavoring elements, and fragrances. It also sees the need to introduce a policy to

intensify the use of raw herbs in the production of essential oils and herbal products.

Product diversification opportunities in the tea sector are to be explored, based on taste

and quality preferences.

Through value addition, processing and vertical product diversification, there is

scope for not just increasing exporting earnings, but also producing more complex products

based on agro-forestry resources. This holds true for both NTIS and Trade Policy products,

as well as other agriculture and forest products that may be identified in future. Not all

products are the same, in that both medicinal plants and essential oils are targeted by

NTIS. However, essential oils (based on medicinal plants) are far more interconnected with

other products than medicinal plants.

On average, processed agro/forestry/food products are more sophisticated and

carry greater potential for structural transformation than do primary commodities. Focusing

on such products that are relatively nearby, Nepal current’s export basket could offer a feasible path to structural transformation without compromising the objectives of mass

employment creation and poverty alleviation. Table 21 compares summary statistics of

PRODY, Path and Strategic Value of processed agro products not currently exported by

Nepal with comparative advantage. However, the products lie at a near-to-medium distance

from Nepal’s current export basket (RCA>1) with those of non-processed agro products,

whether exported by Nepal or not.

To enhance industrialization in the long term, capabilities must be increased to

competitively produce more sophisticated manufactured products, using domestic primaryagro-forest resources. These are highly connected with other products (e.g. perfumery

instead of just essential oils from aromatic plants, Ayurveda medicine from medicinal

plants/herbs, dye from cardamom, diverse paper products based on handmade paper

made of local plants).



63

The strategy of promoting exports of products based on agro-forestry resources is

in itself not wrong, given the country’s resource endowment. But failure to adopt policy

measures to encourage manufacturing of agro-forestry-based products increases the risk

of falling into the commodity-dependence trap. To avoid that trap, success of export

promotion strategies must be measured in terms of progress in exports of commodity-

based manufactured products versus unprocessed commodities. Even when manufacturing

process is limited by the very nature of the product, emphasis must be laid on quality

upgrading and branding to capture higher value. After all, seemingly homogenous

agriculture products (such as tea and coffee) can be vertically differentiated based on their

intrinsic quality and priced accordingly.38 NTIS, Trade Policy and future trade policies and

strategies must be implemented in that direction.

Taking the cash incentive scheme for exports as an example – it was introduced in

2010/11 and continued in 2011/12 with a budget of NPR 300 million (about $4 million). The

budget for the schemeremains mostly unspent due to procedural rigmarole, among other

things. Notably, under the budget exporters are entitled to 2 to 4 per cent of their

convertible currency export earnings based on the rate of value addition.39 The programme

is not confined to NTIS/Trade Policy-identified products only. Overhalf of the firms that were

Table 21. Processed vs. non-processed agricultural products

Processed (with near-to-medium distance from current export basket)

No. of Mean Median Standard Min Max

products Deviation

PRODY 39 11,742.74 11,708.03 4,799.96 3,919.23 24,747.86

Path 39 135.17 135.49 25.22 74.16 186.34

Strategic 39 10,557.23 10,682.18 2,356.89 4,554.49 16,064.73

value

Non-processed

PRODY 59 8,321.57 6,280.79 6,256.56 1,371.40 33,442.19

Path 59 97.31 102.04 36.25 2.89 159.78

Strategic 46 7,424.54 7,304.57 3,100.12 180.99 14,594.91

value


Note: Classification as processed and non-processed broadly follows that adopted by UnitedStates Department of Agriculture, which excludes fish products. HS codes specified byUSDA are converted to SITC codes. Strategic Values for non-processed products arecalculated only for the 46 products that lie in the Open Forest.

38 See, for example, Rollo (2012) for an analysis of the determinants of Tanzanian export prices.

39 It covers only exports to countries other than India. The cash incentive is 2 per cent, 3 per cent or 4 per

cent depending on whether the rate of value addition is 30-50 per cent, 50-80 per cent or above 80 per cent.



64

awarded the incentive (in 2011/12)40 are agro-food exporters. This is most likely because

domestic value addition should be naturally high for agro-food products, including

unprocessed ones (Kharel, 2012). Simplifying the procedures for granting the cash

incentive could increase the utilization of the scheme and many exporters demanded a flat

rate of incentive irrespective of value addition. However, procedural problems had to bebalanced against the need to encourage as much domestic value addition as possible. The

Ministry of Commerce and Supplies suggested providing cash incentive at a flat rate,

irrespective of value addition, or product-specific rates.41

But apart from procedural issues, it is not clear whether provision of the incentive

will actually induce the exporters concerned to export more (ibid.). Addressing critical

domestic supply-side constraints to exports requires provision of facilities in the nature of

public good or service (e.g. strengthened laboratories, quality inputs, research and

extension). Using cash incentive scheme budget to alleviate these constraints in a targeted

manner is an option worth exploring, as exporters/producers are unable to overcome the

constraints individually (ibid.).

Additionally, policymakers should also be mindful of the possibility the scheme may

create a bias against production and export of processed agro-forest products, including

agro-forest-based industrial products (ibid.). This is since exports of agriculture and forest

products in raw form are likely to qualify for 4 per cent cash incentive by virtue of them

being produced/naturally growing within the country.

Another issue related to the incentive regime is the proposed Special EconomicZone (SEZ), where enterprises are entitled to a range of tax and non-tax incentives and

facilities. While the construction of two SEZs is progressing, and the GoN has announced

its plan to develop six more, the SEZ Bill is still pending in parliament. Interestingly, NTIS

2010 calls for the removal (in the draft SEZ Bill) of the 75 per cent export requirement for

enterprises based in the zone. As the primary purpose of establishing SEZs is to promote

exports (as mentioned in Industrial Policy 2010), the proposal to remove the 75 per cent

export requirement could detract from the potential of using SEZs as an instrument to

increase exports. This is in the context of an extremely poor export performance of Nepal.

Furthermore, as the facilities and concessions granted to firms located in the zones entailcost to the government and represent scarce resources to be utilized most judiciously

(Kharel, 2012).

A dedicated special export zone or a special export zone located within an SEZ can

address this issue, but this is not being considered. NTIS 2010 also recommends that

a positive list be replaced with a negative list specifying the types of industries that SEZs

cannot host. However, if exports are to be sustainable and broad-based conferring greater

benefits on the economy, the SEZs should be used as an instrument of encouraging higher

(genuine) sophistication and domestic content of exports (not just increasing the quantum

40 Rijal, Krishna. 2012. “Cash incentives to 57 export-oriented firms,” Arthik Abhiyan, 2 August, p.1 and 5.

41 “Flat rate or commodity-based rate recommended for export incentives”, Republica (online), 11 November

2012.



65

of exports), and the facilities provided therein should be aimed at encouraging the

acquisition of the required production and export capabilities. This would call for some

judicious discriminationon grounds of tradability of goods and services, backward

integration, value addition and degree of transformation in the production process, despite

the associated administrative difficulties.

Technology is a major factor behind export sophistication and, by implication, export

location. Other major economic factors are marketing, logistics and proximity, fragment

ability, information and familiarity, natural resources, infrastructure, and value chain

organization (Lall et al., 2005). Policy factors – such as trade and industrial policies, trading

blocs and trade preferences – also play a role in determining patterns of specialization and

export sophistication, and location (ibid.).

Supply-side constraints (SSCs) affect both the quantum of exports and the type of

goods exported. Critical SSCs traditionally facing Nepal include inadequate infrastructure,low human capital and inadequacy of trade facilitation measures (including the cost of

being landlocked) (see Adhikari, 2011). These should be distinguished from deteriorating

industrial relations, frequent strikes and shutdowns. As well as deteriorating security

situation that have vitiated the overall business/investment climate and brutalized the

manufacturing sector since 2006.

The growth diagnostics framework for identifying the binding constraints to growth,

considers low export sophistication, low export diversification and very limited export

sophistication possibilities (low Open Forest) as “symptoms” of coordination failure andself-discovery externalities (Hausmann et al., 2005 and 2008). However, in the context of

Nepal, these also appear to be symptoms of other constraints – notably those concerning

infrastructure, human capital and finance. This acts to keep possibilities of export

sophistication and diversification at low levels. For example, a producer’s reluctance or

inability to upgrade his capabilities to produce more sophisticated goods – say, moving

from simple processing of aromatic plants to extracting essential oils from them, to

manufacturing perfumery – could be due to lack of resources to purchase the required

machinery or lack of knowledge of such possibilities. This is rather than, say, the concern

that competitors will also follow suit after seeing his success.

Finally, the manufacturing export sector is a part of the national manufacturing

sector, that is characterized by low productivity and whose performance in the recent

decade has been alarmingly poor. As seen in Section 4, despite structural change in the

Nepali economy being growth-enhancing in the recent decade, overall productivity growth

has been low. Productivity growth in manufacturing has been negative. Due to weak

manufacturing performance and the limited capacity of other sectors to absorb surplus

labour, the shift in labour from agriculture to more productive sectors has been slow. Policy

measures aimed at expediting structural transformation of the economy must foremost be

directed at shoring up the ailing manufacturing sector.



66

Conclusion

This paper reports on the investigation into the nature of recent economic growth in

Nepal. It uses some of the analytical tools and insights from recent literature on the

importance of structural transformation as a facilitator of long-term growth. In particular, itexamines the extent Nepal’s recent trade policy initiatives take into account the long-term

potential to assist economic growth. This is through the encouragement of production and

export of goods and services, which could facilitate the shift of economic activities towards

more complex and sophisticated products. In turn, it has the potential to set the economy

on a sustainable long-term growth trajectory.

The analysis shows that while economy-wide structural change – reallocation of

labour across sectors – has contributed positively to productivity growth in Nepal in a recent

decade, overall productivity growth has been low. More so, productivity growth in

manufacturing has been negative, and due to weak manufacturing performance and the

limited capacity of other sectors to absorb surplus labour, the shift in labour from agriculture

to more productive sectors has been slow. To address the root causes of these problems,

attention should be drawn to symptoms of de-industrialization and demand appropriate

policy measures. This would also further expedite structural transformation of Nepal’s

economy.

Although, trade has been recognized as an engine of growth at the policy level,

export performance has been weak, particularly so in the past decade. This is despite

policy measures, over the last two decades, to integrate the Nepali economy into the globaleconomy, mainly through economic liberalization. Nepal’s export performance has been

moribund, in terms of value and growth. More so, its merchandise export basket remains

low in technological sophistication and poorly diversified. Structural transformation, as

measured by the rate of increase in the sophistication of the export basket, has slowed

sharply in the last decade. Further, opportunities for producing and exporting new and more

sophisticated goods (as represented by the export basket) are limited. The expansion of

such opportunities has slowed down significantly. The challenge that Nepal faces on the

export front is two-fold – a) to increase the export earnings and employment by increasing

the export of existing products, given the current industrial and export structures andcapabilities. This is so as to reduce the ever-widening trade imbalance and increase

income and employment levels, and b) to upgrade the industrial and export structures, and

capabilities to be able to produce and export (or expand) the production and export of more

sophisticated products requiring a greater number of capabilities.

Considerations of export potential, socio-economic impact and utilization of

domestic agro-forestry resources underlie the choice of products for priority development

and promotion in the Government of Nepal’s policy documents, notably Trade Policy (TP)

2009 and Nepal Trade Integration Strategy (2010). While Trade Policy 2009 provides theoverall trade-related policy direction, it is NTIS 2010 that is to guide the government’s

trade-related priorities until 2015. NTIS accords top priority to agro-food products. Our

analysis shows that NTIS products differ in their sophistication and potential for structural



67

transformation, but overall they are below-average on these parameters. Iron and steel

products have mostly high-to-medium sophistication and connectedness and above-

average Strategic Value, and most of them lie in the “core” of the product space. However,

agriculture products like tea, lentils, ginger, cardamom and medicinal herbs are at the low

end. Handmade paper, some pashmina and wool products, honey and essential oils offer better prospects. Trade Policy products (that is, products that are exclusive to Trade Policy)

are also overall below-average, but fare better than some NTIS products, particularly when

iron and steel products are excluded from the NTIS product set. Leather goods, paper

products, processed leather, vegetable seeds, towel and non-wood handicraft hold

prospects that are above average, as well as surpassing quite a few NTIS products. There

is some scope to better the prospects of achieving greater export sophistication and

diversification by drawing in some of the products identified by Trade Policy, but excluded

from NTIS.

The choice of products for targeting is limited by the concentration of Nepal’s

effective exportable products in the periphery of the product space, with low sophistication

and low connectedness. In turn, this is responsible for most of the unexploited “nearby”

products being relatively low in sophistication, connectedness and Strategic Value. This

would involve producing and exporting the “nearby” products, which requires capabilities

similar to the existing capabilities. However, there exist some unexploited nearby products

with relatively high potential for further export diversification, which should be considered

while targeting products. Of course, this is subject to the condition that certain minimum

demand, supply-side and competitiveness conditions are met. In addition, some of the

“other” export potential products identified by NTIS, but not accorded priority (such as dairy

products and cement) carry greater potential for structural transformation than the “main”

NTIS products, excluding iron and steel. In future, it is necessary that the structural

transformation dimension also inform trade policymaking and implementation.

Given Nepal’s resource endowment, the strategy of promoting exports of products

based on agro-forestry resources is justified. But unless this is accompanied by policy

measures to encourage manufacturing of agro-forestry-based products, in the long-term

such a strategy increases the risk of falling into the commodity-dependence trap. To avoid

that trap, success of export promotion strategies must be measured in terms of theprogress made in increasing the share of commodity-based manufactured productsover

time. This is due to processed agro/forestry/food products generally being more

sophisticated, and having greater potential for structural transformation than primary

commodities. NTIS, Trade Policy and future trade policies and strategies must be oriented

towards that direction. This ought to be so not only for NTIS products, but also Trade Policy

products and other agriculture and forest products that may be identified in future. Focusing

on such products that are relatively nearby, Nepal current’s export basket could offer a

feasible path to structural transformation without compromising the immediate objectives of

mass employment creation and poverty alleviation. From a long-term perspective, in order to enhance industrialization, capabilities must be increased to competitively produce more

sophisticated manufactured products, using domestic primary agro-forest resources. These

are highly connected with other products (e.g. perfumery instead of just essential oils from



68

aromatic plants; Ayurveda medicine from medicinal plants/herbs; dye from cardamom;

diverse paper products based on handmade paper made from local plants).

Supply-side constraints affect both the quantum (value and volume) and

sophistication of exports. The Government’s latest trade strategy identifies sector-specific

and cross-cutting constraints to exports. It then charts out a course of actions to alleviate

them. In making the effort to alleviate, attention must be paid to product sophistication and

diversification constraints, not just production and exportation of the same type of goods.

Notwithstanding the importance of producing and exporting more sophisticated

products, policymakers must not lose sight of the actual extent (or nature) of value addition

and manufacturing that takes place inside the economy. Sophistication of a final product

alone may shed little light on these aspects. Vertical production diversification and quality

upgrading of existing exports are also important. Furthermore, it must be noted that the

method of product space analysis to assess structural transformation possibilities is basedon exports. It fails to capture the possibilities associated with import-substituting sectors. A

comprehensive view of structural transformation, therefore, must also take into account the

import-competing sectors. These are important as tradable sectors, and full of possibilities

for future exports.



70

Felipe, J., U. Kumar, N. Usui, and A. Abdon. (2010c). “Why Has China Succeeded? And

Why it Will Continue to Do So.” Working Paper No. 611, Levy Economics Institute.

Government of Nepal (GoN). (2009). Trade Policy 2009. Kathmandu: GoN/Ministry of

Commerce and Supplies.

Government of Nepal (GoN). (2010a). Nepal Trade Integration Strategy 2010: Executive

Summary and Action Matrix. Kathmandu: GoN/Ministry of Commerce and Supplies.

Government of Nepal (GoN). (2010b). Industrial Policy 2010. Kathmandu: GoN/Ministry of

Industry.

Government of Nepal (GoN). (2010c). Nepal Trade Integration Strategy 2010: Background

Report. Kathmandu: GoN/Ministry of Commerce and Supplies.

Government of Nepal (GoN). (2011). Three-Year Plan 2010/11-2012/13. Kathmandu: GoN/

National Planning Commission.

Grossman, G.M., and E. Helpman. (1991). “Quality ladders in the theory of growth.” Review

of Economic Studies 58(1): 43-61.

Hausmann, R., and B. Klinger. 2006. “Structural transformation and patterns of comparative

advantage in the product space.” Harvard University Center for International

Development Working Paper No. 128, available at http://www.hks.harvard.edu/

centers/cid/publications/faculty-working-papers/cid-working-paper-no.-128

Hausmann, R., J. Hwang and D. Rodrik. (2007). “What you export matters.” Journal of

Economic Growth 12(1): 1-25.

Hausmann, R., and D. Rodrik. (2003). “Economic development as self-discovery.” Journal

of Development Economics. 72: 603-633.

Hausmann, R., and B. Klinger. (2007). “The structure of the product space and the

evolution of comparative advantage.” Harvard University Center for International

Development Working Paper No. 146 (extended and revised version of No. 128).

Hausmann, R., and C.A. Hidalgo. (2010). “Country diversification, product ubiquity and

economic divergence.” Harvard University Center for International Development

Working Paper No. 201, October.Hausmann, R., B. Klinger and R. Wagner. (2008). “Doing growth diagnostics in practice: A

‘mindbook’. Harvard University Center for International Development Working Paper

No. 177, September.

Hausmann, R., D. Rodrik and A. Velasco. (2005). “Growth diagnostics.” Manuscript, Inter-

American Development Bank.

Helpman, E., and P.R. Krugmann. (1985). Market structure and foreign trade.Cambridge,

MA: MIT Press.

Hidalgo, C.A., and R. Hausmann. (2009). “The building blocks of economic complexity”.PNAS 106(26), 30 June: 10570-10575.

Hidalgo, C.A., and R. Hausmann. (2008). “A network view of economic development.”

Developing Alternatives 12(1): 5-10.



71

Hidalgo, C.A., B. Klinger, A.L. Barabasi and R. Hausmann. (2007). “Product space

conditions the development of nations.” Science 317: 482-487.

Hidalgo, C.A., (2009). “The dynamics of economic complexity and the product space over

a 42 year period.” Harvard University Center for International Development Working

Paper No. 189, December.

Hidalgo, C.A., (2011). Discovering Southern and East Africa’s industrial opportunities.

Economics Policy Paper Series 2011. Washington, D.C.: The German Marshall

Fund of the United States.

Hirschman, A.O., (1958). The strategy of economic development . New Haven, Conn.: Yale

University Press.

ITC. (2007). “Export Potential Assessment in Nepal.” Report prepared by International

Trade Centre, Geneva and Trade and Export Promotion Centre, Kathmandu.

Kaldor, N., (1967). Strategic factors in economic development . Ithaca, NY: New York State

School of Industrial and Labour Relations, Cornell University.

Kharel, P., (2012). Role of commodities in socio-economic development of Nepal. Paper

submitted to United Nations Conference on Trade and Development (UNCTAD),

Geneva, April.

Krugman, P.R., (1979). “Increasing returns, monopolistic competition, and international

trade.” Journal of International Economics 9: 469-479.

Lall, S., (2000). “The technological structure and performance of developing country

manufactured exports, 1995-1998.” Oxford Development Studies, 28(3): 337-369.

Lall, S., J. Weiss and J. Zhang. (2005). “The ‘sophistication’ of exports: A new measure of

product characteristics”. QEH Working Paper Series – QEHWPS 123, January.

Leamer, E., (1984). Sources of international comparative advantage: Theory and evidence.

Cambridge, MA: MIT Press.

Lin, J., and H.J. Chang. (2009). “Should industrial policy in developing countries conform to

comparative advantage or defy it? A debate between Justin Lin and Ha-Joon

Chang.” Development Policy Review 27(5).McMillan, M., and D. Rodrik. 2011. “Globalization, structural change and productivity

growth.” NBER Working Paper Series, w17143, available at www.nber.org

Melitz, M.J., (2003). “The impact of trade on intra-industry reallocations and aggregate

industry productivity.” Econometrica, 71: 1695-1725.

Nepal Rastra Bank, “Recent Macroeconomic Situation”, various issues.

Rodrik, D., (2011a). “The future of economic convergence.” NBER Working Paper 17400,

September, available at www.nber.org/papers/w17400

Rodrik, D., (2011b). “Unconditional convergence.” Centre for Economic Policy Research

Discussion Paper No. 8631, November.



72

Rollo, V., (2012). “Determinants of Tanzanian export prices.” World Bank Policy Research

Working Paper 6225, October.

Romer, P.M., (1990). “Endogenous technological change”. Journal of Political Economy 98

(5, II): 71-102.

Rosenstein-Rodan, P., (1943). “Problems of industrialization of Eastern and South Eastern

Europe.” Economic Journal 53 (June-September): 202-211.

World Trade Organization (WTO). (2012). Trade Policy Review [of Nepal]: Report by the

Secretariat, Geneva.

Xu, B., (2010). “The sophistication of exports: Is China special?” China Economic Review

21: 482-493.



73

Appendix

TableA1.Top30

“nearby”productsindescendingorderofStrategicValue(year2010)

SITC

Product

L

eamergroup

Path

PRODY

PRODY

Pat

h

Contribution

Distance

Strategic

code

tercile

terc

ile

toopen

value

forest

(PRODY*

density)

8932

Sanitaryortoiletart.

La

bour

195.22

13,541.54

medium

high

2,289.36

nearby

15,076.17

ofmaterialsofdiv.58

intensive

6421

Boxes,bags&oth.packing

Fo

restproducts

182.21

14,854.13

medium

high

2,576.47

nearby

14,184.04

con

tainers,ofpaper/papbd

6924

Casks,drums,boxesofiron/Capitalintensive

183.55

12,609.02

medium

high

2,187.44

nearby

14,059.19

steelforpackinggoods

6417

Paper&paperboard,

Fo

restproducts

183.28

10,462.18

medium

high

1,814.01

nearby

13,886.84

cor

rugated,creped,

crin

kledetc.

6794


Capitalintensive

185.55

11,749.49

medium

high

2,065.54

nearby

13,790.18

intheroughstate

8212

Furnitureformedical,

La

bourintensive

176.70

13,331.69

medium

high

2,296.04

nearby

13,489.34

sur

gical,dentaletc.practice

6975

Sanitarywareforindooruse,Capitalintensive

173.04

15,926.95

high

high

2,716.29

nearby

13,319.42

andparts

6973


Capitalintensive

179.19

12,795.86

medium

high

2,324.09

nearby

13,104.04


6517

Yarnofregeneratedfibres,

Capitalintensive

178.03

9,639.35

low

high

1,804.43

nearby

13,042.27

notforretailsale



74

TableA1.

( c o n t i n u e d )

SITC

Product

L

eamergroup

Path

PRODY

PRODY

Pat

h

Contribution

Distance

Strategic

code

tercile

terc

ile

toopen

value

forest

(PRODY*

density)

6560

Tulle,lace,embroidery,

Capitalintensive

171.36

12,836.79

medium

high

2,497.21

nearby

12,899.75

ribbons,&othersmallwares

484

Bakeryproducts(e.g.bread,Cereals,etc.

170.66

14,416.13

medium

high

2,619.15

nearby

12,820.16

bis

cuits,cakes)etc.

7758

Ele

ctro-thermicappliances,

Machinery

167.83

14,429.25

medium

high

2,637.38

nearby

12,548.79

n.e

.s.

620

Sugarconfectioneryand

Tr

opical

165.64

10,445.02

medium

high

1,773.48

nearby

12,542.11

oth

ersugarpreparations

ag

riculture

8124

Lig

htingfixturesandfittings

Capitalintensive

165.62

14,004.12

medium

high

2,567.28

nearby

12,400.53

andparts

6512

Yarnofwooloranimalhair

Capitalintensive

162.09

12,826.38

medium

high

2,278.88

nearby

12,360.81

(includingwooltops)

6536

Fabrics,wovencontain.

Capitalintensive

163.92

10,978.14

medium

high

2,110.75

nearby

12,254.98

85

percentofdiscont.

reg

ener.fibr.

7752

Ho

useholdtype

Machinery

166.96

13,599.29

medium

high

2,490.03

nearby

12,210.56

refrigerators&foodfreezers

8211

Ch

airsandotherseats

La

bourintensive

172.04

10,686.91

medium

high

2,105.14

nearby

12,161.61

andparts



75

TableA1.


SITC

Product

L

eamergroup

Path

PRODY

PRODY

Pat

h

Contribution

Distance

Strategic

code

tercile

terc

ile

toopen

value

forest

(PRODY*

density)

6664

Tab

leware&otherarticles

La

bourintensive

163.76

9,012.59

low

high

1,728.93

nearby

12,012.98

ofporcelainorchina

6732

Bars&rods,ofiron/steel;

Capitalintensive

164.60

7,426.31

low

high

1,317.53

nearby

11,938.29

hollowminingdrillst.

8219

Oth

erfurnitureandparts

La

bourintensive

168.16

11,803.49

medium

high

2,311.17

nearby

11,936.50

6651

Containers,ofglass,

La

bourintensive

167.34

9,161.98

low

high

1,699.10

nearby

11,896.34

use

dforconveyanceor

packing

6129

Oth

erarticlesofleather

Capitalintensive

157.30

10,377.48

medium

high

1,952.18

nearby

11,747.61

orofcomposit.leather

6259

Oth

ertyres,tyrecases,

Capitalintensive

164.39

8,756.15

low

high

1,677.47

nearby

11,671.19

innertubes

6519

Yarnoftext.fibres,n.e.s.,

Capitalintensive

158.30

7,461.56

low

high

1,421.60

nearby

11,666.84

incl.yarnofglassfib.

7731

Insulated,elect.wire,cable,

Machinery

160.92

7,533.37

low

high

1,484.09

nearby

11,656.13

bars,stripandthelike

6416

Buildingboardofwood

Fo

restproducts

157.94

13,516.99

medium

high

2,386.73

nearby

11,635.39

pulporofvegetablefibre

583

Jam

s,fruitjellies,

Tropical

162.12

7,027.30

low

high

1,337.97

nearby

11,493.51

ma

rmalades,

ag

riculture

fruitpuree,cooked



76

TableA1.


SITC

Product

L

eamergroup

Path

PRODY

PRODY

Pat

h

Contribution

Distance

Strategic

code

tercile

terc

ile

toopen

value

forest

(PRODY*

density)

6130

Furskins,tanned/dressed,

Ca

pitalintensive

158.88

16,585.08

high

high

3,271.88

nearby

11,422.45

pieces/cuttingsoffurskin

6349

Wo

od,simplyshaped,n.e.s.Fo

restproducts

152.35

9,271.73

low

high

1,592.64

nearby

11,330.38

S o u r c e :

Auth

or’scalculation,seetext.



77

TableA2.Top30productsindescend

ingorderofStrategicValue(year2010)

SITC

Product

L

eamergroup

Path

PRODY

PRODY

Pat

h

Contribution

Distance

Strategic

code

tercile

tercile

toOpen

Value

Forest

(PRODY*

density)

5335

Colour.preptnsofakind

Chemical

196.55

18,099.09

high

high

2,307.83

faraway

17,717.46

usedinceramic,enamelli.

6632

Naturalorartificialabrasive

La

bourintensive

194.49

18,839.34

high

high

2,423.08

faraway

17,615.70

powderorgrain

7267

Otherprintingmach.

Machinery

174.30

21,054.27

high

high

2,200.40

faraway

17,527.86

for

usesancillarytoprinting

7492

Taps,cocks,valvesetc.

Machinery

189.17

17,626.42

high

high

2,209.09

faraway

17,306.65

for

pipes,tanks,vatsetc.

7269

Partsofthemachines

Machinery

176.80

20,564.04

high

high

2,337.46

faraway

17,207.22

of726.31,726.4-,726.7-

6418

Paper&paperboard,

Fo

restproducts

189.02

19,418.90

high

high

2,430.43

faraway

17,197.90

impregnat.coat.

surface-coloured

7849

Otherparts&accessories

Machinery

193.79

17,158.54

high

high

2,166.88

faraway

17,185.51

ofmotorvehicles

7429

Partsofthepumps&

Machinery

176.21

18,779.22

high

high

2,127.31

faraway

17,088.95

liq.

elevatorsof742—

7439

Partsofthemachines

Machinery

181.99

16,500.97

high

high

1,975.90

faraway

17,027.87

of743.5-,743.6-

6996

Miscellaneousarticles

Capitalintensive

207.69

15,704.52

high

high

2,418.90

middle

17,006.43

ofbasemetal



78

TableA2.


SITC

Product

L

eamergroup

Path

PRODY

PRODY

Pat

h

Contribution

Distance

Strategic

code

tercile

terc

ile

toopen

value

forest

(PRODY*

density)

6785

Tube&pipefittings

Capitalintensive

207.19

15,323.85

high

high

2,385.44

middle

16,929.91

(joints,elbows)ofiron/steel

7449

Partsofthemachinery

Machinery

199.53

16,022.39

high

high

2,239.96

middle

16,886.23

of744.2-

7139

Partsofint.comb.piston

Machinery

194.09

16,920.96

high

high

2,314.52

middle

16,883.36

enginesof713.2-/3-/8-

8935

Art.ofelectriclighting

La

bourintensive

198.18

15,538.38

high

high

2,341.37

middle

16,768.43

ofmaterialsofdiv.58

6210

Ma

terialsofrubber

Capitalintensive

198.81

15,040.43

high

high

2,219.33

middle

16,665.21

(e.g.pastes,plates,

she

ets,etc.)

7493

Tra

nsmissionshafts,cranks,Machinery

171.15

17,263.96

high

high

1,855.28

faraway

16,647.64

bearinghousingsetc.

7132

Int.combustionpiston

Machinery

167.49

17,148.64

high

high

1,787.76

faraway

16,576.36

enginesforpropellingveh.

7428

Oth

erpumpsforliquids&

Machinery

174.12

16,345.75

high

high

1,842.59

faraway

16,505.85

liqu

idelevators

7212

Harvesting&treshing

Machinery

183.58

18,719.55

high

high

2,339.47

faraway

16,488.80

ma

chineryandparts

7413

Ind

.&lab.furnacesand

Machinery

178.57

17,122.77

high

high

2,119.06

faraway

16,388.38

ove

nsandparts



79

TableA2.


SITC

Product

L

eamergroup

Path

PRODY

PRODY

Pat

h

Contribution

Distance

Strategic

code

tercile

terc

ile

toopen

value

forest

(PRODY*

density)

7234

Constructionandmining

Machinery

171.63

14,324.30

medium

high

1,632.89

faraway

16,386.08

ma

chinery,n.e.s.

7188

Engines&motors,n.e.s.

Machinery

186.10

17,844.67

high

high

2,358.53

faraway

16,367.01

suc

haswaterturbinesetc.

7368

Wo

rkholders,self-opening

Machinery

172.69

18,167.42

high

high

2,001.33

faraway

16,365.09

dieheads&toolholders

7442

Lifting,handling,loading

Machinery

181.05

17,010.07

high

high

2,138.61

faraway

16,332.11

ma

ch.conveyors

7219

Agric.mach.&appliances,

Machinery

179.47

20,207.11

high

high

2,559.10

faraway

16,322.78

n.e

.s.andparts

7783

Ele

ctr.equip.forinternal

Machinery

170.56

15,925.70

high

high

1,846.94

faraway

16,299.60

com

bustionengines,parts

7436

Filtering&purifyingmach.

Machinery

175.48

17,104.32

high

high

2,080.48

faraway

16,283.89

for

liquids&gases

2331

Synth.rubb.lat.;synth.

Tropical

175.08

9,595.75

low

high

1,130.67

faraway

16,266.49

rub

b.facticederiv.fromoils

ag

riculture

6953

Oth

ertoolsforusein

Ca

pitalintensive

189.71

14,733.62

medium

high

2,200.75

middle

16,174.98

the

hand

6991

Loc

ksmithswares,safes,

Ca

pitalintensive

189.13

16,414.08

high

high

2,416.89

middle

16,155.87

stro

ngroomsofbasemetal

S o u r c e :

Auth




80

TableA3.NTISproducts

SITC

Product

NTIS

Leamer

A

lsoin

PRODY

Path

PRODY

Path

Strategic

Distance

code

group

group

Trade

tercile

tercile

Value

Policy?

483

Maca

roni,spaghetti

Insantnoodle

Cereals,etc.

No

13,574.83

91.50

medium

low

6,955.17

ands

imilarproducts

542

Bean

s,peas,lentils&

Lentils

Tropical

Yes

4,339.02

105.04

low

low

6,996.98

other

leguminous

agriculture

vegetables

616

Naturalhoney

Natu

ralhoney

Tropical

Yes

9,330.78

133.60

low

medium

9,379.04

nearby

agriculture

741

Tea

Tea

Tropical

Yes

1,371.40

86.31

low

low

5,889.44

agriculture

752

Spice

s(exceptpepper

Card

amom/

Tropical

Yes

3,207.49

104.23

low

low

6,647.25

andp

imento)

ginger

agriculture

2820

Wasteandscrapmetal

Ironandsteel

Raw

No

6,290.71

134.81

low

medium

9,881.07

nearby

ofironorsteel

products

materials

2924

Plants,seeds,fruitused

Medicinalherbs

Animal

Yes

3,909.24

98.21

low

low

6,088.21

inperfumery,pharmacy

products

5513

Essentialoils,concretes&

Esse

ntialoils

Chemical

Yes

5,244.98

127.73

low

medium

8,986.14

absolutes;resinoids

6412

Printingpaper&writing

Handmade

Forest

Yes

20,158.88

171.62

high

high

14,714.97

middle

paper,inrollsorsheets

pape

r

products

6583

Trave

llingrugsand

Pash

mina

Capital

Yes

8,175.97

144.30

low

medium

9,502.38

blank

ets,notknitted/

products

intensive

croch

eted



81

6712

Pigiron,castironand

Irona

ndsteel

Capital

No

6,647.54

117.97

low

medium

8,769.1

0

middle

spiegeleisen,inpigs,blocks

produ

cts

intensive

6713

Irono

rsteelpowders,

Irona

ndsteel

Capital

No

10,002.32

105.91

medium

low

8,844.9

8

faraway

shoto

rsponge

produ

cts

intensive

6716

Ferro-alloys

Irona

ndsteel

Capital

No

13,126.03

121.97

medium

medium

8,271.4

8

nearby

produ

cts

intensive

6724

Puddledbarsandpilings;

Irona

ndsteel

Capital

No

11,361.39

135.62

medium

medium

10,881.8

6

middle

ingots

,blocks,lumpsetc.

produ

cts

intensive

6725

Bloom

s,billets,slabs&

Irona

ndsteel

Capital

No

6,710.47

132.79

low

medium

9,731.9

6

middle

sheet

barsofironorsteel

produ

cts

intensive

6727

Irono

rsteelcoilsfor

Irona

ndsteel

Capital

No

8,405.86

145.82

low

medium

11,284.7

6

middle

re-rolling

produ

cts

intensive

6731

Wirerodofironorsteel

Irona

ndsteel

Capital

No

8,747.24

170.97

low

high

13,280.4

9

middle

produ

cts

intensive

6732

Bars&rods,ofiron/steel;

Irona

ndsteel

Capital

No

7,426.31

164.60

low

high

11,938.2

9

nearby

hollow

miningdrillst.

produ

cts

intensive

6733

Angle

s,shapes&

Irona

ndsteel

Capital

No

12,322.83

182.66

medium

high

14,331.1

9

middle

sectio

ns&sheetpiling,

produ

cts

intensive

ofiron/st.

6744

Sheets&plates,

Irona

ndsteel

Capital

No

10,024.73

179.40

medium

high

13,791.2

7

middle

rolled

>4.75mmofiron/steelprodu

cts

intensive

TableA3. (

c o n t i n u e d )

SITC

Product

NTIS

Leamer

Alsoin

PRODY

Path

PRODY

Path

Strategic

Distance

code

group

group

T

rade

tercile

tercile

Value

Po

licy?



82

6745

Shee

ts&plates,rld.

Ironandsteel

Capital

No

12,127.31

180.25

medium

high

14,778.09

middle

thickns.3mmto4,75mm

products

intensive

irn/stl.

6746

Shee

ts&plates,rolled;

Ironandsteel

Capital

No

12,105.81

169.18

medium

high

13,445.87

middle

thicknessoflessthan3mm

products

intensive

6747

Tinne

dsheetsandplates,

Ironandsteel

Capital

No

13,147.33

157.56

medium

high

13,037.03

middle

ofste

el

products

intensive

6749

Othersheetsandplates,

Ironandsteel

Capital

No

14,363.66

157.16

medium

high

13,073.81

ofironorsteel,worked

products

intensive

6760

Rails

andrail waytrack

Ironandsteel

Capital

No

13,937.68

159.82

medium

high

14,089.85

faraway

constructionmaterial

products

intensive

6770

Iron/steelwire/wheth/not

Ironandsteel

Capital

No

11,951.71

181.76

medium

high

13,921.51

coate

d,butnotinsulated

products

intensive

6781

Tubesandpipes,

Ironandsteel

Capital

No

12,931.14

145.67

medium

medium

12,291.38

middle

ofcastiron

products

intensive

6782

seam

lesstubesandpipes;

Ironandsteel

Capital

No

10,939.27

167.16

medium

high

13,713.96

middle

blank

sfortubes&pipes

products

intensive

6783

Othertubesandpipes,

Ironandsteel

Capital

No

13,034.23

181.49

medium

high

14,125.68

ofironorsteel

products

intensive

6785

Tube

&pipefittings

Ironandsteel

Capital

No

15,323.85

207.19

high

high

16,929.91

middle

(joints,elbows)ofiron/steel

products

intensive

6793

Steel&ironforgings&

Ironandsteel

Capital

No

14,865.17

189.09

medium

high

15,521.88

middle

stampings,inroughstate

products

intensive

TableA3.


SITC

Product

NTIS

Leamer

A

lsoin

PRODY

Path

PRODY

Path

Strateg

ic

Distance

code

group

group

Trade

tercile

tercile

Value

Policy?



83

6794


Ironandsteel

Capital

No

11,749.49

185.55

medium

high

13,790.1

8

nearby

inthe

roughstate

products

intensive

6911

Struc

tures&partsof

Ironandsteel

Capital

No

13,657.89

193.44

medium

high

14,782.3

4

struc.;iron/steel;plates

products

intensive

6921

Rese

rvoirs,tanks,

Ironandsteel

Capital

No

13,472.55

203.59

medium

high

15,887.4

6

middle

vatsandsimilarcontainers

products

intensive

6924

Casks,drums,boxesof

Ironandsteel

Capital

No

12,609.02

183.55

medium

high

14,059.1

9

nearby

iron/s

teelforpackinggoods

products

intensive

6931

Stran

dedwire,cables,

Ironandsteel

Capital

No

11,323.75

171.04

medium

high

12,745.5

2

corda

gesandthelike

products

intensive

6932

Wire,twistedhoopfor

Ironandsteel

Capital

No

8,837.42

124.40

low

medium

9,590.9

7

middle

fencingofironorsteel

products

intensive

6935

Gauz

e,cloth,grillofiron

Ironandsteel

Capital

No

12,741.14

167.43

medium

high

13,644.2

1

steel

orcopper

products

intensive

6940

Nails,screws,nuts,

Ironandsteel

Capital

No

15,655.73

165.65

high

high

15,288.8

9

faraway

bolts

etc.ofiron,steel,

products

intensive

copper

6973


Ironandsteel

Capital

No

12,795.86

179.19

medium

high

13,104.0

4

nearby


products

intensive

6974

Art.c

ommonlyusedfordom.Ironandsteel

Capital

No

10,507.25

123.06

medium

medium

8,657.6

0

purpo

ses,potscourers

products

intensive

TableA3.


SITC

Product

NTIS

Leamer

A

lsoin

PRODY

Path

PRODY

Path

Strateg

ic

Distance

code

group

group

Trade

tercile

tercile

Value

Policy?



84

6975

Sanitarywareforindoor

Ironandsteel

Capital

No

15,926.95

173.04

high

high

13,319.42

nearby

use,andparts

products

intensive

6992

Chainandpartsthereof,

Ironandsteel

Capital

No

14,492.18

182.00

medium

high

15,367.24

middle

ofironorsteel

products

intensive

6993

Pins&needles,fittings,

Ironandsteel

Capital

No

12,664.65

154.09

medium

high

14,002.18

middle

base

metalbeads,etc.

products

intensive

6994

Sprin

gs&leavesfor

Ironandsteel

Capital

No

16,376.32

141.15

high

medium

13,215.29

faraway

springs,ofiron/steel/copper

products

intensive

6997

Articlesofironorsteel,

Ironandsteel

Capital

No

14,708.04

187.79

medium

high

15,607.88

middle

n.e.s.

products

intensive

8121

Boilers&radiatorsfor

Ironandsteel

Capital

No

17,100.74

180.83

high

high

14,976.54

middle

centralheating

products

intensive

8421

Overcoatsandother

Woolproducts

Labour

Yes

6,894.52

137.94

low

medium

8,584.60

coats

,men,s

intensive

8424

Jackets,blazersoftextile

Woolproducts

Labour

Yes

8,109.45

128.03

low

medium

8,033.28

fabric

s

intensive

8431

Coatsandjacketsof

Woolproducts

Labour

Yes

7,420.63

132.83

low

medium

8,168.26

textilefabrics

intensive

8432

Suits

&costumes,

Woolproducts

Labour

Yes

7,141.74

139.00

low

medium

8,683.87

women’s,oftextilefabrics

intensive

8435

Blousesoftextilefabrics

Woolproducts

Labour

Yes

6,388.06

115.49

low

low

6,757.63

intensive

TableA3.


SITC

Product

NTIS

Leamer

A

lsoin

PRODY

Path

PRODY

Path

Strateg

ic

Distance

code

group

group

Trade

tercile

tercile

Value

Policy?



85

TableA3.


SITC

Product

NTIS

Leamer

A

lsoin

PRODY

Path

PRODY

Path

Strateg

ic

Distance

code

group

group

Trade

tercile

tercile

Value

Policy?

8439

Otheroutergarmentsof

Woolproducts

Labour

Yes

6,398.12

115.05

low

low

6,890.7

0

nearby

textile

fabrics

intensive

8451

Jerse

ys,pull-overs,

Woolproducts

Labour

Yes

7,176.82

112.53

low

low

6,643.6

1

nearby

twinsets,cardigans,knitted

intensive

8452

Dresses,skirts,suitsetc.,

Woolproducts

Labour

Yes

9,148.28

126.88

low

medium

7,852.6

4

knitte

dorcrocheted

intensive

8459

Otheroutergarments&

Woolproducts

Labour

Yes

7,563.43

111.61

low

low

6,709.4

6

clothing,knitted

intensive

8471

Clothingaccessories

Pash

mina

Labour

Yes

4,814.31

118.11

low

medium

7,869.4

0

oftex

tilefabrics

products

intensive

8472

Clothingaccessories,

Wool

Labour

Yes

6,232.33

137.57

low

medium

8,800.1

0

knitte

dorcrochetem,

products

intensive

n.e.s.

8484

Head

gearandfittings

Woolproducts

Labour

Yes

7,167.00

100.35

low

low

7,160.6

2

thereof,n.e.s.

intensive

8972

Imitationjewellery

Silve

rjewellery

Labour

Yes

13,733.01

115.95

medium

low

8,966.1

7

nearby

intensive

8973

Jewelleryofgold,silver

Silve

rjewellery

Labour

Yes

8,006.96

98.70

low

low

6,488.8

6

nearby

orpla

tinum

intensive

S o u r c e :

Auth


N o t e :

Blan

k“distance”cellsmeanthattheproductisnotintheOpenForest.



86

TableA4.Trade

Policyproducts

SITC

Product

TradePolicy

Leamer

PRODY

Path

PRODY

Path

Strategic

Distance

code

group

group

tercile

tercile

Value

541

Potatoes,freshorchilled,excl.

Fresh

Tropical

10,573.11

134.72

medium

medium

10,036.92

nearby

swe

etpotatoes

vegetables

agriculture

545

Oth

erfreshorchilledvegetables

Fresh

Tropical

5,164.89

111.65

low

low

7,318.41

nearby

vegetables

agriculture

546

Veg

etables,frozenorin

Fresh

Tropical

9,090.29

135.91

low

medium

9,193.83

nearby

tem

porarypreservative

vegetables

agriculture

571

Ora

nges,mandarins,

Orange

Tropical

8,169.07

103.77

low

low

7,168.19

nearby

clementinesandothercitrus

agriculture

711

Coffee,whetherornotroasted

Coffee

Tropical

1,812.91

64.62

low

low

3,792.50

nearby

orfreedofcaffeine

agriculture

2614

Silk

wormcocoonssuitabl.

Silk

Cereals,

1,576.46

63.90

low

low

4,929.39

forreeling&silkwaste

products

etc.

2925

See

ds,fruit&spores,nes,

Vegetable

Animal

12,201.77

121.42

medium

medium

9,272.25

nearby

ofa

kindusedforsowing

seeds

products

2926

Bulbs,tubers&rhizomes

Floriculture

Animal

10,317.39

91.81

medium

low

7,231.63

nearby

offloweringoroffoliage

products

2927

Cutflowersandfoliage

Floriculture

Animal

4,076.19

74.46

low

low

5,330.02

nearby

products

6113

Calfleather

Processed

Capital

6,191.60

128.08

low

medium

8,713.38

nearby

leather

intensive

6114

Lea

therofotherbovinecattle

Processed

Capital

8,413.08

126.28

low

medium

8,600.82

and

equineleather

leather

intensive



87

TableA4.


SITC

Product

TradePolicy

Leamer

PRODY

Path

PRODY

Path

Strategic

Distance

code

group

group

tercile

tercile

Value

6115

She

epandlambskinleather

Processed

Capital

3,253.27

97.00

low

low

6,176.41

nearby

leather

intensive

6116

Lea

therofotherhidesorskins

Processed

Capital

2,503.84

109.55

low

low

7,065.60

leather

intensive

6118

Lea

ther,speciallydressed

Processed

Capital

10,891.87

140.79

medium

medium

11,288.56

nearby

orfinised

leather

intensive

6121

Articlesofleatherorof

Leather

Capital

16,321.95

176.45

high

high

15,526.73

middle

com

positionleather

goods

intensive

6122

Sad

dleryandharness,

Leather

Capital

11,242.10

133.53

medium

medium

10,404.13

nearby

ora

nymaterialforanimals

goods

intensive

6129

Oth

erarticlesofleatheror

Leather

Capital

10,377.48

157.30

medium

high

11,747.61

nearby

ofc

omposit.leather

goods

intensive

6354

Manufacturesofwoodfor

Handicraft

Forest

5,536.48

115.20

low

low

7,392.37

dom

estic/decorativeuse

(wood)

products

6411

Newsprint

Paper

Forest

20,806.53

110.68

hig

h

low

9,624.18

faraway

products

products

6413

Kra

ftpaperandpaperboard,

Paper

Forest

17,950.03

161.78

high

high

13,188.05

middle

inrollsorsheets

products

products

6415

Pap

erandpaperboard,inrolls

Paper

Forest

18,526.86

173.58

high

high

14,270.82

middle

ors

heets,n.e.s.

products

products

6417

Pap

er&paperboard,

Paper

Forest

10,462.18

183.28

medium

high

13,886.84

nearby

corrugated,creped,crinkledetc.

products

products



88

TableA4.


SITC

Product

TradePolicy

Leamer

PRODY

Path

PRODY

Path

Strategic

Distance

code

group

group

tercile

tercile

Value

6418

Pap

er&paperboard,

Paper

Forest

19,418.90

189.02

hig

h

high

17,197.90

faraway

imp

regnat.coat.surface-coloured

products

products

6421

Box

es,bags&oth.packing

Paper

Forest

14,854.13

182.21

medium

high

14,184.04

nearby

con

tainers,ofpaper/papbd

products

products

6422

Writingblocks,envelopes,

Paper

Forest

15,115.28

183.81

hig

h

high

14,852.15

etc.correspondencecards

products

products

6423

Registers,exercisebooks,

Paper

Forest

10,073.57

141.56

medium

medium

10,518.95

notebooks,etc.

products

products

6424

Pap

erandpaperboard,cutto

Paper

Forest

15,599.70

190.87

hig

h

high

16,131.59

middle

size

orshape,n.e.s.

products

products

6428

Art.ofpaperpulp,paper,

Paper

Forest

14,071.13

181.26

medium

high

14,596.65

middle

pap

erboard,cellu.wadding

products

products

6511

Silk

yarn&yarnspunfrom

Silk

Capital

8,838.03

89.21

low

low

7,088.16

nearby

noil/othersilkwaste

products

intensive

6521

Cottonfabrics,woven,

Towel

Capital

3,658.41

116.84

low

medium

7,289.40

nearby

unb

leached,notmercerized

intensive

6522

Cottonfabrics,woven,bleach.

Towel

Capital

8,295.30

144.96

low

medium

9,896.62

merceriz.dyed,printed

intensive

6541

Fab

rics,woven,ofsilk,ofnoil

Silk

Capital

8,087.67

82.29

low

low

6,733.89

nearby

oro

therwastesilk

products

intensive

6549

Fab

rics,woven,n.e.s.

Towel

Capital

10,757.39

114.32

medium

low

9,014.21

intensive



89

TableA4.


SITC

Product

TradePolicy

Leamer

PRODY

Path

PRODY

Path

Strategic

Distance

code

group

group

tercile

tercile

Value

6560

Tulle,lace,embroidery,ribbons,

Towel

Capital

12,836.79

171.36

me

dium

high

12,899.75

nearby

&othersmallwares

intensive

6577

Wadding.textil.fabricsforuse

Towel

Capital

15,767.07

176.93

hig

h

high

15,901.78

faraway

inm

achinery/plant

intensive

6592

Car

pets,carpetingandrugs,

Woolen

Capital

1,728.10

74.04

low

low

4,824.88

kno

tted

carpet

intensive

6666

Statuettes&oth.ornaments,

Handicraft

Labour

10,323.48

137.44

me

dium

medium

10,025.22

&articlesofadornment

(non-wood)

intensive

6673

Oth

.precious&semi-precious

Gemsand

Labour

4,970.44

85.74

low

low

5,814.00

nearby

stones,unwork.cutetc.

jewellery

intensive

6978

Hou

seholdappliances,

Handicraft

Capital

6,854.79

106.81

low

low

8,211.38

dec

orativeart.,mirrorsetc.

(non-wood)

intensive

8310

Travelgoods,handbags,

Leather

Labour

7,843.86

97.19

low

low

6,914.61

nearby

brie

f-cases,purses,sheaths

goods

intensive

8422

Suits,men’s,oftextilefabrics

RMG

Labour

8,585.71

140.88

low

medium

9,078.77

intensive

8423

Trousers,breechesetc.

RMG

Labour

5,758.79

110.52

low

low

6,656.95

oftextilefabrics

intensive

8429

Oth

eroutergarmentsof

RMG

Labour

6,372.41

114.18

low

low

6,636.24

nearby

textilefabrics

intensive

8433

Dre

sses,women’s,

RMG

Labour

5,718.43

125.13

low

medium

7,560.81

oftextilefabrics

intensive



90

TableA4.


SITC

Product

TradePolicy

Leamer

PRODY

Path

PRODY

Path

Strategic

Distance

code

group

group

tercile

tercile

Value

8434

Skirts,women’s,oftextilefabrics

RMG

Labour

6,825.56

136.04

low

medium

8,342.09

intensive

8441

Shirts,men’s,oftextilefabrics

RMG

Labour

5,337.75

112.22

low

low

6,669.40

intensive

8442

Und

ergarments,excl.shirts,

RMG

Labour

4,889.04

104.95

low

low

6,126.59

oftextilefabrics

intensive

8443

Und

ergarments,women,s,

RMG

Labour

5,518.67

91.12

low

low

5,258.91

oftextilefabrics

intensive

8462

Und

ergarments,knittedofcotton

RMG

Labour

5,315.72

108.39

low

low

6,340.14

nearby

intensive

8463

Und

ergarments,knitted,

RMG

Labour

4,862.42

118.33

low

medium

7,132.28

nearby

ofs

yntheticfibres

intensive

8465

Cor

sets,brassieres,

RMG

Labour

6,992.34

118.85

low

medium

7,469.71

nearby

suspendresandthelike

intensive

8481

Art.

ofapparel&clothing

Leather

Labour

5,032.63

129.12

low

medium

8,946.49

nearby

accessories,ofleather

goods

intensive

8974

Oth

erarticlesofpreciousmetal

Gemsand

Labour

9,461.95

104.98

low

low

10,904.11

faraway

jewellery

intensive

8999

Manufacturedgoods,n.e.s.

Leather

Labour

5,485.91

72.81

low

low

5,350.79

nearby

goods

intensive

S o u r c e :

Auth


N o t e :

Blan

k“distance”cellsmeanthattheproductisnotintheOpenForest.



91

II. Logistics performance and trade: An analysisof India’s trade in intermediates with

Bangladesh and Thailand

By Prabir De and Amrita Saha

Introduction

Logistics is an important determinant in sustaining a country’s (or a region’s)

competitive advantage.42 Its contribution to growth, economic integration and poverty

reduction is well known. Improvements in logistical services help countries produce more

sophisticated products and encourage a more dynamic export (import) diversificationprocess. In turn, this contributes to improvements in an economy’s growth and

development. During the past decade there have been noticeable developments in logistic

services due to technology. However, there is evidence of a rising gap between the Least

Developing Countries (LDC) and the developing economies. This is in terms of quality of

services in Asia and the Pacific region.43

Logistic services involve the process of planning, implementing and controlling

efficient and cost-effective flow, storage of raw materials, in-process inventory, finished

goods and related information, from a point of origin to the point of consumption(destination) to meet customer requirements.44 Production processes and tasks are

becoming increasingly fragmented across national borders. As a result, time-sensitive

logistic services, along with information and communication technology (ICT), are the key in

facilitating production networks spanning borders.45 In other words, logistic services play

a catalytic role by ensuring just-in-time delivery of goods and services, either as inputs to

production process or as final output networks. Efficiency in logistical services contributes

42 Refer to for example, World Bank (2012), Planning Commission (2011), to mention a few.

43 Based on a worldwide survey of operators on the ground – such as global freight forwarders and expresscarriers – the Logistics Performance Index (LPI) of the World Bank measures the logistics “friendliness” of

155 countries. It helps countries identify the challenges and opportunities they face in their trade logistics

performance and what they can do to improve. For example refer to World Bank (2012). Appendix 1 presents

the global ranks of selected Asia-Pacific countries for the year 2012. The contrast is, while Singapore and

Hong Kong, China occupy the first and second global rank in LPI, countries like Mongolia, Myanmar,

Cambodia, and Lao People’s Democratic Republic fall in the bottom group in LPI. This shows wide intra-

regional variations in logistics performance.

44 There is no clear consensus on definition of logistics. In literature, it overlaps in many cases with

transportation, even though there is a clear difference between the two. In most ASEAN and South Asian

countries, there is still a lack of understanding of what makes up logistics and how a logistics policy should be

developed. Logistics development policy frequently becomes just a transport investment infrastructure plan, but

logistics is much more than just transport infrastructure. Developing a national logistics policy requires a holistic

approach that encompasses traders, service providers, infrastructure, and rules and regulations. Refer, for

example, Hollweg and Wong (2009), Sourdin and Pomfret (2012).

45 This is what Baldwin termed as “the 2nd unbundling”. The 2nd unbundling is the international division of

labour in terms of production processes and tasks. Refer to Baldwin (2011)



92

to the expansion of trade and production networks within or across countries, as well as

building their productive capacities in networked countries (Ando and Kimura, 2009; Jones

and Kierzkowski, 2001; Kimura, 2012). Efficiency in logistic services is also dependent, to

a large extent, on ‘Behind the Border’ measures of government policy and regulation, which

are driven by efficiency and equity concerns. It is therefore important to have a regionallogistics sector policy to facilitate the trade and production linkages across Asia’s borders.46

In this study, our objective is to empirically explore the role of logistics in enhancing

production linkages through trade in intermediate products. Two case studies of trade in

intermediate goods are undertaken: (i) India’s export of textile yarn to Bangladesh; and

(ii) India’s import of air conditioning equipment from Thailand. Here, yarn and air

conditioning equipment are selected owing to a steady rise in trade of these two products,

which are facilitated by regional and bilateral FTAs.47 Bangladesh buys yarn from India, and

India buys air conditioning equipment from Thailand. These can facilitate the development

of integrated production networks.

The remaining part of the study is organized as follows. Section 2 undertakes

a review of the existing literature and identifies the gaps. Section 3 presents the data and

methodology. Section 4 discusses some stylized facts about India’s trade with Bangladesh

and Thailand in yarn and air conditioning equipment. Also discussed are the overall trends

in intra-industry trade between them. Section 5 undertakes an assessment of logistics in

Asia-Pacific countries including Bangladesh, India and Thailand. Section 6 presents the

major analytical findings, and Section 7 concludes the paper.

A. Logistics services and production networks:Literature review

Manufactured goods are the largest and most rapidly growing portion of world trade.

Studies show that a country’s global competitiveness is improved by more efficient supply

chains and better access to logistics services. More so, it creates the conditions for

mutually beneficial production linkages across borders.48 As production is increasingly

shared across borders, simplification of trade processes and procedures would helpimprove the time and costs associated with logistics, thereby improving export

competitiveness.49

On the other hand, higher trade costs and service inefficiency may discourage

fragmentation of production. While reduction in transaction costs through efficient logistics

46 This is also not to deny that framing a regional logistics sector policy has been slow in South and

Southeast Asia, compared to national logistics sector policies adopted by several developing countries in

recent years, Refer, for example, to Findlay (2009), Sourdin and Pomfret (2009).

47 Preferential tariff reductions are given under, for example, SAFTA in case of India – Bangladesh trade, and

ASEAN-India FTA and India-Thailand FTA in case of India – Thailand trade.

48 Refer to, for example, Arnold (2010)

49 See, for example, Duval and Utoktham (2011), ESCAP (2011), to mention a few.



93

services, liberalization of trade in services and investment policy assists cross-border

production linkages. Here, liberalization of trade in services is defined as the demand of

any services that arises directly from trade itself (in this context international trade), and

some examples of trade in services are transportation, communication, insurance, banking

etc. According to Deardorff (2001), this liberalization permits rationalization of serviceactivities along the lines of comparative advantage. He cites the case of US-Mexico

cross-border transportation services as an example. Before the North American Free Trade

Agreement (NAFTA), Mexican truckers were not allowed to enter US territory and vice

versa. Thus, if goods are to be shipped from Mexico to the US, then Mexican trucks would

carry the goods up to Mexico-US border checkpoint, unloaded from Mexican registered

truck, reloaded on to a US registered truck and carried to the destination. As a result, the

consignment faced a number of transaction costs in the form of time, customs delay, and

regulatory costs and so on. After NAFTA, liberalization of such transportation services has

allowed a consignment to be shipped in a Mexican truck up from its origin in Mexico to itsUS destination. This method has reduced transport costs and time, as well as helping to

reduce the final price of goods. Hummels (2007) also elaborates on this in relation to sea

and air transportation.

Logistical services constitute an essential part of international trade that usually

involves progression in communication, transportation, logistics, finance, etc.

Communication and transport are not only vital intermediate inputs to international trade,

but can also be final exports. The competitiveness of the manufacturing firms in open

economies is determined in part by access to low-cost and high-quality producer services

(Francois and Hoekman, 2010). Efficiency in logistic services can thereby generate benefits

for merchandise trade flows by directly reducing the associated transaction costs, and

indirectly improving competitiveness of firms (Mattoo et al., 2001; Deardoff, 2001). Hesse

and Rodrigue (2004) discuss improvement in logistics as including four core elements;

traditional transport costs, organization of the supply chain, transactional and physical

environments in which freight distribution evolves. This enables private firms to expand

their opportunities more efficiently. Within this, a product (or its component) inputs cross

international borders several times during the process of its production, in accordance with

related economic incentives. In such a scenario, service link costs can have a multiplicative

effect on the total cost of producing a final product (Hiratsuka, 2008; Kimura and

Kobayashi, 2009).

Owing to diversity in the range of logistical services facilitating trade, efficient

regulation of logistic services is sector specific. In telecommunications this may refer to pro-

competitive regulation, while in the financial service sector it will concern prudential

regulation (Mattoo, Rathindran and Subramanian, 2001). In transport, these may be

reflected in lower freight, documentation and administrative costs in customs procedures.

As supply chains become more complex, and expand over a larger space, logistical

requirements become more sophisticated and demanding, increasing pressure onunderlying infrastructure (Brooks, 2008; 2010).



94

As noted in Kimura and Kobayashi (2009), the key to attracting fragmented

production blocs is to (i) improve locational advantages by, for example, developing special

economic zones (SEZs). This with at least an improved local level investment climate, and

(ii) reduce the cost of service links that connect remotely located production blocs by

improving trade and transport facilitation. Figure 1 presents graphical links betweenproduction blocks and service links. In fragmentation of production, an efficient and

improved service link is important for expansion of production networks across a region.

Figure 1. Production blocks and logistics service links

Source: Kimura and Kobayashi (2009).

East Asia has recorded high intra-regional trade shares owing, in particular, to

rapidly expanding intra-regional trade in parts and components. This is particularly so for Association of Southeast Asian Nations (ASEAN), where production network exports is

over 60 per cent of total manufacturing exports in the last decade (Athukorala, 2010). Their

progress in recent decades is attributed, at least in part, to technological changes in service

industries, where there has been more rapid and effective transportation and

communication. The ICT revolution and time-sensitive logistics infrastructure is considered

to have played a major role in the “2nd unbundling” of manufacturing production in Asia,

which began in the 1980s (Kimura, 2012).50 For example, a positive relationship has

developed between competition and privatization in telecom sectors for 12 developing East

Asian countries. Mishra et al. (2011) suggests that increasing sophistication in serviceexports carries important implications for countries in Asia. Particularly so for countries

50 The term “2nd unbundling” describes international division of labour in terms of production processes and

tasks was coined by Baldwin (2011).

Before fragmentation

After fragmentation

Large integrated factory

PB : Production blocks

SL : Service links

SL

SL

SL

SL

SL

PB

PB

PB

PB

PB



95

stuck in a middle-income trap (Malaysia, Viet Nam etc.), or for those that wish to sustain

their rapid growth (India, Sri Lanka, etc.). But the World Bank’s Logistics Performance

Index (LPI) indicates that the systems of logistics in South Asia must be developed further

to successfully meet the challenges of product fragmentation. However, rigorous evidence

on the impact of improved logistical services on cross-border production linkages andproduction networks is sparse.

This is designed to make a contribution to this under-researched area.

B. Data and methodology

The authors aim to answer the following research questions in this study.

• How do we measure logistical service performance?

• Does logistical service performance play a catalytic role in expanding the flow

of intermediates between India and Bangladesh, and India and Thailand?

• What is the causality between logistical service performance and industrial

fragmentation?

We aim to understand how changes in logistic efficiency affect changes in import

demand among sectors that generate production networks. To do so, a quantitative

analysis of this relationship will be conducted by specifying labour model based on the

following constant elasticity of substitution (CES) equation.51

U i = ( Σ j

λ j x

) (1)

where i and j are importing and exporting countries, respectively, θ = σ / (1 - σ). We treat λ

as a quality shifter specific to exporter j . In other words, it represents the number of unique

varieties being produced by exporter j .

We write the import demand for a product as follows:

qij = E

i ( ) t ij -

σ(2)

where qijis the value of import of i from j , t is trade cost component which captures logistics

efficiency. E is real expenditures on a product (expenditures divided by the price level),

which we do not observe but proxy it by country’s GDP.52 Similarly, λ /p are not really

51 Labour Substituting scarce factors of production by relatively more abundant ones is a key element of

economic efficiency and a driving force of economic growth. A measure of that force is the elasticity of

substitution between capital and labour, which translates into a constant elasticity of trade with respect to trade

cost.

52 The reason is that if all goods are consumed as a constant fraction of GDP and price levels do not vary,

but we do not see the expenditure shares or the price levels. In particular, the main way that international

production sharing shows up here is that E varies a lot across countries as a function of what they are

producing – a country makes lot of cars it demands an unusually large amount of car parts and components.

1 / θ

θ

σ

p j

λ j



96

observable due to poor quality of measures of p, and it is also contaminated by quality

differences.53 Unfortunately, prices net of quality differences and quality itself are not

available. We proceed as follows:

First, we take logs and use a vector of importer and exporter fixed effects. We get

equations (3) and (4) below:

In qij = In E

i + σ In ( ) – σ In t

ij (3)

In qij = A

i + A

j – σ In t

ij (4)

Second, we replace tijby z

ij , which is the logistics performance index (LPI). We write

the trade cost vector as follows:

In qij = A

i + A

j – σ In Z

ij (5)

Since our purpose is to assess the impact of LPI on trade over time, two years are

considered, namely, 2000 and 2010. We rewrite the equation (2) as follows:

= (6)

By taking logs, we get:

In = In ( ) + σ In – σ In ( ) (7)

We incorporate importer and exporter fixed effects, and rewrite it as follows:

In = Ai + A j – σ In ( ) (8)

Now, controlling for other exogenous variables, we rewrite the equation (8) as

follows:

In = Ai + A

j – σ In ( ) – σ In X '

ij + ε

ij (9)

where i and j are importing and exporting countries. We use country dummy (= 1 when i is

importer (exporter), and 0 otherwise). The parameters to be estimated are denoted by σ ,

and ε ij is the error term.

53 For example, a high price for a product may reflect higher production costs, or it may just reflect quality

differences.

qij 2000

qij 2010 ( )E i 2010

( )E i 2000

z ij 2010 -σ

σ

p j 2010

λ j 2010

p j 2000

λ j 2000

σ

z ij 2000 -σ

qij 2000

qij 2010 E i 2010

E i 2000 ( ) p j 2000

λ j 2000

p j 2010

λ j 2010

qij 2000

qij 2010

z ij 2010

z ij 2000

z ij 2010

z ij 2000

qij 2000

qij 2010 z ij 2010

z ij 2000

p j

λ j



97

Whether the trade between two or more economies will rise and be facilitated

towards fragmentation will depend on the potential of intra-industry trade between them.

Following Mikic and Gilbert (2007), we attempt to assess the magnitude and emerging

trend of the intensity of intra-industry trade (IIT). Here, data is sourced from COMTRADE.

We measure logistic services across countries by generating an index based on

a selected set of indicators. This uses multi-dimensional factor analysis (principal

component analysis). Data is sourced from WDI.

To understand the relationship between logistic service efficiency and trade in

intermediates, a panel data regression is carried out with equation (9) as the baseline.

Co-integration technique is also used to assess the direction of causality between logistics

performance and trade.

We use 2000 and 2010 data to empirically estimate this relationship for Bangladesh’s import of yarn from India, and import of air conditioning equipment by India

from Thailand. We consider only the stage of yarn supply to Bangladesh’s ready-made

garment producers and India’s import of air conditioning equipment from Thailand. We do

not go further to explore in detail other characteristics of the supply chain, such as whether

they comprise a specific production network as discussed in recent literature.54

C. India’s trade with Bangladesh and Thailand

India’s trade with Bangladesh and Thailand is influenced by its Free Trade

Agreements (FTAs), such as SAFTA in the case of Bangladesh, and the India-Thai

FTA – ASEAN FTA, in the case of Thailand.55 India’s yarn exports to Bangladesh and

air-conditioning equipment (ACE) imports from Thailand are major components of their

bilateral trade (Table 1).

India’s imports of ACE from Thailand grew substantially during the last decade,

comprising 12.35 per cent of India’s total imports from Thailand in 2010. ACEs are part of

India-Thailand EHS, where India has offered tariff concessions and reduced the customs

duty to zero. India’s export of yarn to Bangladesh accounted for about 35 per cent of India’stotal exports to Bangladesh.

We now study these trade flows disaggregated in terms of intermediates, capital

and consumption goods, using the Broad Economic Categories (BEC) classification.56 In

2010, India’s total imports of “Transport Equipment and Parts and Accessories thereof”

54 The usual caveat is that one needs to link the supplier of inputs with users of the product in a backward

linkage framework.

55 SAFTA was implemented among eight South Asian countries on 1 July 2006, whereas India – Thai EHS

was implemented on 1 March 2004, and India – ASEAN FTA came in force on 1 January 2010.

56 Refer Appendix 2 for the BEC Codes and corresponding BEC-HS correspondences. Details on this

methodology can be accessed at: http://www.icrier.org/pdf/amrita_saha.pdf. A limitation of this consists of the

fact that a single intermediate maybe an input for several final goods. It only traces the evidence of possibilities

of production networks. This can be useful when supported by surveys with firms involved in these networks.



98

Table 1. India’s imports of ACE from Thailand and export of yarn to Bangladesh:

2000-2010

Year India’s Import India’s Total India’s Export India’s Total

of ACE from Import from of Yarn to Export to

Thailand Thailand Bangladesh Bangladesh

(Million USD)

2000 5.42 335.38 209.12 860.33

2001 22.71 404.38 201.36 1,000.63

2002 42.50 390.02 183.00 1,132.54

2003 45.11 551.54 208.34 1,599.55

2004 35.31 777.38 292.38 1,624.82

2005 112.00 1,125.16 308.49 1,656.05

2006 161.30 1,612.10 358.13 1,636.98

2007 214.17 2,162.16 346.59 2,594.56

2008 251.67 2,567.24 805.60 2,574.66

2009 257.12 2,683.95 469.66 2,181.10

2010 470.61 3,810.14 1,070.86 3,021.79

CAGR (%) 56.25 27.51 17.74 13.39

Source: Based on COMTRADE. For corresponding HS codes, please refer Appendix 2.

Figure 2. Trends in trade shares: India’s export of yarn to Bangladesh

and import of ACE from Thailand

1.62

5.61

8.18

12.35

24.31

20.12

16.16

13.03

18.63

21.88

13.36

31.29

21.53

35.44

9.589.809.9110.01

9.95

4.54

10.90

17.99

0

5

10

15

20

25

30

35

40

2 0 0 0

2 0 0 1

2 0 0 2

2 0 0 3

2 0 0 4

2 0 0 5

2 0 0 6

2 0 0 7

2 0 0 8

2 0 0 9

2 0 1 0

P e r c e n t

Import share Export share



99

from Thailand amount to $482 million. Matched with the corresponding HS codes for ACE,

it is noted that more than 97 per cent of total transport equipment imported from Thailand in

2010 comprised of ACEs. In terms of the BEC, these can be classified under intermediate

goods used in manufacturing capital or consumption goods. These may be used in

producing several consumption goods classified under BEC-522 i.e. the non-industrialtransport equipment and capital goods under BEC-521, as well as industrial transport

equipment.

In the same year, India exported globally $2.2 billion worth of industrial transport

equipment and $0.8 billion worth of non-industrial transport equipment. It is also noted that

India’s imports of ACE from Thailand seem to be primarily driven by Japanese MNEs.

These may indicate early stages of production networks involving Japan, Thailand and

India. However, this has not been investigated in detail.

Similarly, while some Indian firms involved in yarn manufacture appear to be alsoinvolved in spinning, weaving and finishing of textiles, we do not have adequate data to

confirm the existence of strong production networks. Yarn (cotton and polyester), as well as

fabric (mainly denim), are almost exclusively exported by India to Bangladesh through road

transport (by trucks).57 This suggests that logistical performance may have a major impact

on cross-border production linkages work involving India and Bangladesh.

1. Intra-industry trade (IIT) and vertical fragmentation of production

The IIT index measures the degree of overlap between imports and exports in thesame commodity category, with a value of 1 indicating pure intra-industry trade and a value

of 0 indicating pure inter-industry trade.58

Table 2 presents the common set of traded goods between India and Bangladesh,

for which relatively high IIT index scores are observed.59 The estimated scores indicate that

IIT index levels are higher in manufactured products than in primary products, reflecting the

greater role of economies of scale in the production of those products.

The IIT scores suggest that there may be production-sharing opportunities, in

a static sense, in 11 products with varying potential (table 3). This potential varies from thetextile and clothing sector (most concentration) to iron and steel (least concentration).

Whereas, electrical machinery and equipment, and mechanical appliances occupy the

middle portion (medium concentration) of the value chain. The index scores also indicate

that there are only two sectors in which intra-industry trade accounts for a moderate share

between India and Bangladesh. This is namely textile and clothing, and electrical

57 There are some shipments from India’s western part to Bangladesh by ocean.

58 Before calculating IIT, data coordinates at HS nomenclature H2 were matched for both the countries. The

traditional way to measure the degree of intra-industry trade is the Grubel-Lloyd Index (G-L Index). For further

details of IIT, please refer to Mikic and Gilbert (2007, p. 76).

59 Appendix 3 presents the calculated IIT scores.



100

Table 2. Intra-Industry trade index (2007): Common set of products at 6-digit HS

HS Code Product IIT India IIT Bangladesh

230220 RICE BRAN OIL 0.935 0.836

721550 BARS & RODS OTHRTHN FREE-CUTNG 0.923 0.421STL NT FRTHR WRKD THN COLD

FRMD/COLD FINSHD

850720 OTHER LEAD-ACID ACCUMULATORS 0.922 0.557

600622 OTHR KNITED OR CROCHETD FBRCS 0.771 0.929

OF COTTON, DYED

960719 OTHER SLIDE FASTENERS 0.770 0.719

610510 MEN’S/BOYS’ SHIRTS OF COTTON 0.758 0.819

621790 PARTS OF GARMENTS/CLOTHNG 0.729 0.463

ACCESSORIES848390 PARTS OF TRANSMISSION SHAFTS, 0.703 0.778

CRANKS, BEARING HOUSINGS,

GEARS OR CLUTCH

854419 WINDING WIRES OF OTHR METLS/ 0.505 0.633

SUBSTANCES

620319 SUITS OF OTHER TEXTILE MATERIALS 0.486 0.704

521211 OTHR UNBLCHED WOVEN FABRICS OF 0.417 0.770

COTTON WEIGHING NOT MORE

THAN 200 G/M2Note: IIT index was calculated for bilateral trade between India and Bangladesh.

Table 3. IIT in textile and clothing sector, 2010 (Exporter – India,

Importer – Bangladesh)

HS code Product IIT

5608 Knotted netting of twine, cordage or rope; made up fishing nets and other 0.14made up nets, of textile materials

5208 Woven fabrics of cotton, containing 85% or more by weight of cotton, 0.20weighing not more than 200 g/m2

5211 Woven fabrics of cotton, containing less than 85% by weight of cotton, 0.24

mixed mainly or solely with man-made fibres, weighing more than 200 g/m2

5408 Woven fabrics of artificial filament yarn, including woven fabrics obtained 0.32

from materials of heading No. 54.05

5210 Woven fabrics of cotton, containing less than 85% by weight of cotton, mixed 0.33

mainly or solely with man-made fibres, weighing not more than 200 g/m2

5402 Synthetic filament yarn (other than sewing thread), not put up for retail sale, 0.41

including synthetic monofilament of less than 67 decitex

5403 Artificial filament yarn (other than sewing thread), not put up for retail sale, 0.42

including artificial monofilament of less than 67 decitex

5204 Cotton sewing thread, whether or not put up for retail sale 0.48

Source: Calculated using Tradesift, University of Sussex.



101

machinery and mechanical appliances sectors at the 6-digit HS level. In the category of

textile and clothing – cotton sewing thread (HS 5204), artificial filament yarn (HS 5403) and

synthetic filament yarn (HS 5402) have relatively higher IIT scores. In other sectors, intra-

industry trade is small or negligible.

In the case of India’s import of ACE from Thailand, we find relatively higher and

rising IIT index score in air, vacuum pumps, compressors, ventilating fans, etc. (HS 8414).

This increased from 0.344 in 2000 to 0.409 in 2010 with a peak of 0.590 in 2007 (Table 4).

In sharp contrast, IIT index scores of air conditioning equipment, machinery (HS 8415) and

compression-ignition engines (diesel, etc.) (HS 8408) are very low.

Table 4. IIT in air conditioning equipment (Importer – India, Exporter – Thailand)

Compression-ignition Air, vacuum pumps, Air conditioning

engines (diesel, etc.), compressors, equipment, machinery

(HS 8408) ventilating fans, etc. (HS 8415)

(HS 8414)

2000 0.548 0.344 0.031

2001 0.007 0.214 0.013

2002 0.017 0.145 0.020

2003 0.044 0.247 0.031

2004 0.173 0.061 0.051

2005 0.019 0.205 0.064

2006 0.002 0.276 0.013

2007 0.008 0.590 0.003

2008 0.029 0.563 0.003

2009 0.036 0.448 0.001

2010 0.023 0.409 0.008

Source: Calculated using Tradesift, University of Sussex.

To identify the vertical IIT, the indices at a high disaggregated level (HS 6) are

compared with those at a low disaggregated level (HS 2). IIT indices that are low at HS 6

and high at HS 2 are a necessary (although not sufficient) condition for the existence of

vertical trade. This is because they suggest that the countries trade different products in the

same sector. The usual caveat is that when the IIT index is observed to be low at HS 6 but

high at HS 2, one should check on case-by-case basis whether the different products are

differentiated as final products or as parts and components. This is against final products to

minimise aggregation bias. Tables 5 and 6 present the vertical IIT trade potential between

India and Bangladesh, and between India and Thailand, respectively. The textile and

clothing sector appears to offer significant vertical trade opportunities. This is most



102

Table 5. Vertical trade potential between India and Bangladesh*

Reporter Partner HS 2 Commodity (HS 2) IIT IIT** Potential

(HS 2) (HS 6) (HS 2 – HS 6)

India Bangladesh 03 Fish, crustacean, mollusc, 0.970 0.787 0.183

and others

India Bangladesh 09 Coffee, tea, MATN, and 0.801 0.560 0.241

spices

Bangladesh India 03 Fish, crustacean, mollusc, 0.200 0.003 0.197

and others

Bangladesh India 08 Edible fruits and nuts 0.180 0.001 0.179

Bangladesh India 14 Vegetable plaiting materials 0.770 0.012 0.758

Bangladesh India 19 PREP. of cereal, flour, 0.160 0.012 0.149

starch, and milk

Bangladesh India 25 Salt, sulphur, earth, stone, 0.830 0.140 0.691

and plastering materials

Bangladesh India 31 Fertilizers 0.950 0.194 0.756

Bangladesh India 33 Essential oils, resinoids, 0.800 0.476 0.324

perfumery, and cosmetics

Bangladesh India 39 Plastics and articles 0.440 0.326 0.114

thereof

Bangladesh India 53 Other vegetable textile 0.020 0.000 0.020

fibres

Bangladesh India 54 Man-made filaments 0.330 0.019 0.311

Bangladesh India 55 Man-made staple fibres 0.530 0.288 0.242

Bangladesh India 56 Wadding, felt, and 0.690 0.041 0.650

nonwoven yarnsBangladesh India 63 Other made-up textile 0.310 0.235 0.075

articles

Bangladesh India 84 Nuclear reactors, boilers, 0.980 0.277 0.703

parts

Bangladesh India 87 Vehicles of railway, 0.080 0.051 0.029

tramway roll-stock

Notes: *IIT indices are calculated for bilateral trade between India and Bangladesh at HS 2nomenclature. **Average of multiple products at HS 6.

importantly felt in the wadding and nonwoven yarns (HS 56) category. However, the vertical

IIT potential in the case of ACE, at present, seems lower than in some of the other sectors.



103

D. Measuring logistics performance

In this section we briefly summarize the methodology and data sources for

constructing a logistics performance index (LPI) covering 20 Asia-Pacific countries, and the

results.

There are several aspects of logistics which complement each other, such as

telecommunication, transport, financial infrastructure and human resource quality. While

these indicators correlate among themselves in some cases, none of them adequately

capture the overall logistics performance. A country may have a very good network of roads

but poor telecommunication infrastructure, for example. Therefore, the statistical techniqueof principal component analysis (PCA) is helpful in constructing a unique single index

based on information across different variables that reflect different aspects of

infrastructure.

PCA finds linear combinations of the original variables to construct the principal

components (or factors) with a variance greater than any single original variable.

LPI it = Σ W

jt X

jit (10)

where LPIit = Logistics Performance Index of the i-th country (20 countries) in t-th time

(namely, 2000 to 2010), W jt = weight of the j-th aspect of logistics in t-th time, and X

jit = value

of the j-th aspect of logistics for the i-th country in t-th time point.

Table 6. Vertical trade potential between India and Thailand*

Reporter Partner HS 2 Commodity (HS 2) IIT IIT** Potential

(HS 2) (HS 6) (HS 2 – HS 6)

India Thailand 84 Nuclear reactors, boilers, 0.33 0.20 0.13

machinery, etc.

India Thailand 48 Paper & paperboard, 0.99 0.13 0.86

articles of pulp, paper etc.

India Thailand 15 Animal, vegetable fats 0.98 0.11 0.87

and oils, cleavage

products etc.

India Thailand 64 Footwear, gaiters and 0.97 0.11 0.86

the like, parts thereof

India Thailand 51 Wool, animal hair, 0.97 0.05 0.92horsehair yarn and fabric

etc.

India Thailand 87 Vehicles other than 0.97 0.21 0.76

railway, tramway

Notes: *IIT indices are calculated for bilateral trade between India and Thailand at HS 2nomenclature. **Average of multiple products at HS 6.



104

The variables used in the construction of the composite index and their measurements, are

as follows (variables are normalized for economy size).

1. Transportation

The following four indicators have been employed for capturing the availability and

quality of transport infrastructure; (i) air transport is captured with the help of passengers

carried per 1,000 population and air freight taken per 1,000 population, (ii) road

infrastructure is captured by the length of roads network per 100 square kilometre (cm2.) of

surface area, and percentage share of paved roads, (iii) railway infrastructure is captured

Table 7. LPI scores and ranks

Sr. No. Country 2000 2005 2010

Score Rank Score Rank Score Rank

1 Australia 5.143 6 5.334 6 5.487 6

2 Bangladesh 1.269 17 1.476 17 2.130 17

3 Cambodia 1.014 20 1.204 20 2.081 19

4 China 2.489 9 3.383 9 4.213 9

5 Hong Kong, China 8.299 2 9.730 2 10.418 1

6 India 1.776 14 1.993 13 2.882 13

7 Indonesia 2.168 11 2.310 11 3.665 10

8 Japan 5.463 5 5.495 5 6.080 5

9 Korea, Rep. of 5.923 3 5.929 3 7.011 3

10 Lao PDR 1.223 19 1.276 19 2.121 18

11 Malaysia 3.699 7 4.410 7 5.255 7

12 Mongolia 1.545 15 1.730 15 2.313 15

13 Myanmar 1.234 18 1.312 18 1.543 20

14 New Zealand 5.843 4 5.895 4 6.454 4

15 Pakistan 1.312 16 1.603 16 2.289 16

16 Philippines 1.865 12 2.121 12 3.150 12

17 Singapore 10.082 1 10.121 1 10.402 2

18 Sri Lanka 2.354 10 2.523 10 3.571 11

19 Thailand 3.314 8 3.736 8 4.498 8

20 Viet Nam 1.821 13 1.867 14 2.843 14

Spearman rank

correlation 0.992* 0.995* 0.985*

coefficient (2000-2005) (2005-2010) (2000-2010)

*Significant at 1%



105

through length of railway lines per 100 sq km of surface area, and (iv) port infrastructure is

captured by container port traffic per 10,000 population.

2. Information and communication technology

ICT infrastructure is measured with teledensity and density of internet users. Total

number of telephones lines per 1,000 inhabitants is a measure of teledensity. Number of

internet users per 1,000 inhabitants measures IT penetration in logistics.

3. Financial services

Domestic credit provided to the private sector (logistic service providers) by the

banking sector (as per cent of GDP) is employed as a measure of financial infrastructure.

4. Human resource quality

Adult literacy rate is taken as a measure of human resource quality.

The data sources include various issues of World Development Indicators of the

World Bank.

Appendix 4 provides the detailed list of these variables, while Appendix 5 presents

the factor loadings, estimated through PCA. Weights are found to be robust, as factor

loadings for each year explain about 58 to 65 per cent of the observation.

LPI scores and ranks for the 20 countries in 2000, 2005 and 2010 are computed

following the methodology outlined above, and summarized in Table 7. Patterns emerging

from Table 7 are as expected. Note that:

First, Asia and the Pacific are comprised of a heterogeneous group characterized

by wide gaps in logistics performance. Relatively richer economies occupy the top positions

in LPI, whereas the LDCs are at the bottom. For example, Myanmar, Cambodia, Lao

People’s Democratic Republic and Bangladesh occupy the bottom ranks in logistics

performance. Other developing countries occupy the middle portion of the ladder. In

general, the rankings in logistics attainment seem to relate to levels of development.

Second, among the 20 Asia-Pacific countries, six (Bangladesh, Cambodia, Hong

Kong, China, Lao People’s Democratic Republic, Mongolia and Viet Nam) have improved

their ranks between 2000 and 2010. Among those showing improvement, the most

impressive was Viet Nam which moved up 5 places (from 13 to 8). Myanmar moved down,

while seven countries remained in the same positions. The logistics gap between the

relatively developed and the LDCs in Asia and the Pacific region appears to have widened

during this period.



106

E. Does Improvement in Logistical Services Lead to Higher Trade in Intermediate Goods?

Equation (9) was estimated for both India’s export of yarn and imports of ACE as

the dependent variable, to represent external and internal factors that influence trade in

production-networked goods across borders. This estimation is with LPI, for both partner

and reporter countries, and a set of control variables, the exchange rate (er), population

(pop), manufacturing value added (mva), GDP and per capita consumption of electricity

(pce).The panel data model considers a set of 19 Asia-Pacific countries.60 Data was

sourced from WDI.

As there is a strong correlation between GDP and trade, there will a definite

problem if both of these variables are taken together. Hence, the regression models reflects

attempt to avoid the obvious multi-collinearity problem. Also, the data structure shows

non-linearity so that double log regressions give better results than non-transformedvariable-based regressions. Variables being in natural logarithms, estimated coefficients

show CES elasticity. The elasticity is useful as an indicator of the effect of trade barriers on

trade volumes. The estimated baseline results are presented in table 8.

Table 8. Baseline regression (OLS): Fixed effect model

Traditional FEM Traditional FEM

Variable India’s Export of Yarn India’s Import of ACE

ln_export ln_export ln_import ln_import

ln_lpi_r 1.130* 0.266* -2.634*** 0.146*

(0.632) (0.213) (0.993) (0.151)

ln_lpi_p -1.079 0.778* 2.260*** 0.860*

(0.936) (0.661) (0.682) (0.668)

ln_er -0.321*** -1.191 -0.00188 -1.715

(0.0756) (1.904) (0.0719) (2.17)

ln_pce -1.612*** -0.217* -0.933*** -0.145

(0.277) (1.412) (0.219) (1.172)ln_mva 0.194 1.918 6.942*** 2.542

(0.225) (1.689) (0.274) (2.982)

ln_gdp 1.442*** 0.215 0.859*** 0.226

(0.136) (1.327) (0.113) (0.915)

Observations 209 209 209 209

R-squared 0.358 0.952 0.844 0.937

Country effect No Yes No Yes

Year effect No Yes No YesNotes: Robust standard errors in parentheses

*** p <0.01, ** p <0.05, * p <0.1

60 We took all the countries listed in table 7, except Brunei Darussalam. This is due to data limitation.



107

Most variables have expected signs, and adjusted R-squared values range from

0.36 to 0.96.

Note that in all regressions the classical linear regression is dominated by fixed-

effect model. Hence, the ordinary regression results reported in table 8 are not statistically

tenable, particularly in the case of India’s export of yarn.61 Robustness is improved in the

case of fixed-effect model, explaining about 95 per cent of the variations in observation for

export of yarn and 94 per cent for import of ACE.

Baseline regressions suggest that logistical performance and trade in the selected

intermediate goods are positively associated. With these being equal, the improvement of

logistics would lead to an increase in trade. Coefficients of LPI have positive signs in FEM

for both reporting country as well as partner country.

Controlling for country fixed-effects, the estimated elasticity indicates that a 10 per cent improvement in logistical performance in India increases its export of yarn to

Bangladesh by about 3 per cent. Whereas, improvement of logistics, to the same extent, in

Bangladesh increases India’s export of yarn to Bangladesh by almost 8 per cent.

In the case of India’s import of ACE from Thailand, estimated elasticity indicates that

10 per cent improvement in India’s logistics may increase India’s imports by 9 per cent.

Whereas, a 10 per cent improvement in logistics in Thailand may increase its exports of

ACE to India by about 1.5 per cent.

1. Robustness checks

The relationships described above cannot be interpreted as causal until the

possibility of endogeneity has been ruled out in the baseline regressions. To address this

issue, a dynamic GMM estimator (system-GMM) – also known as Arellano-Bover/Blundell-

Bond linear dynamic panel-data estimation – was used to analyse changes across

countries and over time.62 The estimator also effectively deals with reverse causality by

including lagged dependent variables to account for the persistence of the inequality and/or

trade openness indicators.63

61 Selection of model, whether a random-effect or a fixed-effect regression, was done based on Hausmann

test.

62 First introduced by Arellano and Bond (1991).

63 Following Arellano and Bover (1995), and Blundell and Bond (1998), a system-GMM was taken in place of

a difference-GMM. Arellano and Bover (1995) and Blundell and Bond (1998) revealed a potential weakness of

the difference-GMM estimator. They showed that lagged levels can be poor instruments for first-differenced

variables, particularly if the variables are persistent. In their modification of the estimator, they suggested the

inclusion of lagged levels along lagged differences. In contrast to the original difference-GMM, they termed this

the expanded estimator system-GMM.



108

One of the main advantages of the system-GMM estimator is that it does not

require any external instruments other than the variables already included in the dataset. It

uses lagged levels and differences between two periods as instruments for current valuesof the endogenous variable, together with external instruments. More importantly, the

estimator does not use lagged levels or differences by itself for the estimation. Instead, it

employs them as instruments to explain variations in infrastructure development. This

approach ensures that all information will be used efficiently, and that focus is placed on the

impact of regressors (such as trade) on logistics, and not vice versa.

Also, the Arellano-Bond estimates, presented in table 9, remove the weak

instrumental variables and poor efficiency problems, as they utilize more moment

conditions. Table 9 provides system-GMM estimates when the dependent variable is Indianexport of yarn and India’s import of ACE interchangeably. The Wald chi2 statistics indicate

the estimated results are robust and statistically significant. To test the appropriateness of

the instruments used, the Sargan J-statistics of over-identifying restrictions in table 9 is

used. The Sragan J-statistics show that the applied instruments are valid. The Arellano-

Table 9. Arellano-Bover dynamic panel-data estimation (System GMM)

DV = ln_export Coefficient SE DV = ln_import Coefficient SE

ln_export L1 0.239* 0.069 ln_import L1 0.107* 0.030

ln_export L2 0.044 0.083 Ln_import L2 0.015 0.060

ln_lpi_p 0.980*** 0.346 ln_lpi_r 0.584** 0.357

ln_lpi_r 0.654** 0.264 ln_lpi_p 0.168 0.021

ln_er -0.257 0.242 ln_er -0.109 0.118

ln_pce -1.044** 0.434 ln_pce 0.500 0.758

ln_mva 0.095 0.601 ln_mva 0.836 0.846

ln_gdp 1.533*** 0.173 ln_gdp 1.812** 0.395

Wald chi2 2,112.95 2,956.59

(Prob > chi2) (0.00) (0.00)

Sargan test, 2.71 1.63

chi2 (Prob > chi2) (0.342) (0.265)

Arellano-Bond test 1, 0.004 0.003

Prob > z

Arellano-Bond test 2, 0.893 0.675

Prob > z

Instruments 60 60

Observations 171 171

Notes: Dynamic panel counts White period instrument weighting matrix, White period standarderrors and co-variance (d.f. corrected). The estimation uses orthogonal deviation. L1 and L2equal lags 1 and 2, respectively. SE stands for standard errors. *** p <0.01, ** p <0.05,* p <0.1



109

Bond tests for serial correlation support the model specification. If the model is well

specified, we expect to reject the null of no autocorrelation of the first order (AB1), and

accept the hypothesis of no autocorrelation of the second order (AB2). It is apparent that

past exports determine, to some extent, the present level of exports (first period lagged

export is statistically significant). However, logistical performance has a strong influence onthe export of yarn or import of ACE over time. In support of the previous findings (table 8),

system-GMM estimates suggest persistence of export (import), since the initial level of

export (import) appears to be an important instrument in the evolution of production-

networked trade over space and time. Thus, the results of system-GMM support the static

panel result.

Therefore, we conclude that improvements in logistical services are associated with

increased trade in yarn between India and Bangladesh, and in ACE between India and

Thailand. However, to ascertain the causation between logistical performance and trade,

we need to look at the causality.

2. Co-integration and causality

Table 10 presents the results of the Im, Pesaran and Shin (IPS) panel unit root test

at level. IPS test is usually applied for heterogeneous panel to test the series for the

presence of a unit root.64 We found that the null hypothesis of having panel unit root is

generally rejected in all but two variables at level form and various lag lengths. Only two of

the variables are non-stationary on the basis of the IPS test. The results of the panel unit

root tests confirm that the two variables are non-stationary at level. Table 10 also presentsthe results of the tests at first difference for IPS test. It is observed that for all the series the

null hypothesis of unit root test is now rejected at 95 per cent critical value (1 per cent

level). Hence, based on IPS test, there is strong evidence that all the series are integrated

of order one, denoted I (1).

Table 10. Im, Pesaran and Shin (IPS) panel unit root test

(Period: 2000-2010)

Variable Level 1st

Difference

Export of yarn 4.3469

Import of ace 4.1241

lpi_p -0.878 -8.3574

lpi_r -1.2862

gdp 11.1182

mva 0.1723 -8.3376

pce 1.7857

er 3.1842

64 Appendix 6 presents the basic equations of IPS.



110

Next, we tested for co-integration using the four panel co-integration tests

developed by Westerlund (2007).65 The underlying idea is to test for the absence of co-

integration by determining whether the individual panel members are error correcting. This

is to investigate whether long-run steady state or co-integration exist among the variables.

Since the variables are found to be integrated in the same order I (1), we continue with thepanel co-integration tests carried out for constant plus time trend. The postulated

relationship between the variables allows for a linear time trend. The results are in

tables 11(a) and table 11(b). Results strongly reject the hypothesis that the series are not

co-integrated, thereby showing existence of a long-run relationship among the relevant

variables.

Table 11(a). India’s exports of yarn to Bangladesh:

Westerlund panel co-integration test

(Period: 2000-2010)

Statistic Value z-value p-value

gdp (partner)

Gt -4.963 -14.149 0

Ga -21.352 -6.195 0

Pt -18.876 -11.259 0

Pa -85.544 -55.89 0

mva (1st diff)

Gt -11.421 -49.2 0

Ga -23.013 -7.284 0

Pt -33.568 -28.372 0

Pa -34.816 -18.872 0

er

Gt -6.299 -21.399 0

Ga -13.448 -1.016 0

Pt 25.939 40.94 0

Pa 5.263 10.374 0 lpi_p (1st diff)

Gt -12.438 -54.722 0

Ga -21.123 -6.045 0

Pt -0.59 10.039 0

Pa -2.456 4.741 0

lpi _r

Gt -4.587 -12.108 0

Ga -26.657 -9.671 0

Pt -8.86 0.407 0

Pa 19.693 -7.837 0

65 Appendix 7 presents the basic equations of Westerlund.



111

Table 11(b). India’s exports of yarn to Bangladesh:

Westerlund panel co-integration test

(Period: 2000-2010)

Statistic Value z-value p-value

gdp (partner )

Gt -6.076 -20.189 0

Ga -24.845 -8.484 0

Pt -20.045 -12.621 0

Pa -24.635 -11.443 0

mva (1st diff)

Gt -2.128 1.235 0.892

Ga 280.156 191.38 1

Pt -9.967 -0.882 0.189

Pa -28.718 14.423 0

er

Gt -4.728 12.872 0

Ga 1.556 8.816 1

Pt -13.451 -4.94 0

Pa -20.965 -8.765 0

lpi_p (1st diff)

Gt -2.477 -0.656 0.256Ga -118.824 -70.068 0

Pt -8.297 1.063 0.856

Pa -15.949 -5.105 0

lpi_r

Gt -5.87 -19.071 0

Ga -130.242 -77.55 0

Pt -11.751 -2.961 0

Pa -23.23 -10.42 0



112

Finally, we test for causality based on the Granger causality framework.66 By

estimating an equation in which Y is regressed on lagged values of Y and lagged values of

an additional variable X, we can evaluate the null hypothesis that X does not Granger

cause Y. If one or more of the lagged values of X is significant, we are able to reject the null

hypothesis that X does not Granger cause Y. The test results presented in table 12 indicate

a two-way causality between LPI and trade. This, in combination with previous results,

indicates that improvement in logistics in either trading partner would increase trade and

vice versa. This is despite the magnitudes of the effects being different between the trading

partners.

Summary and implications

This is a study of the impact of improved logistic services on India’ trade in two

important intermediate goods – yarn exports to Bangladesh and ACE imports from

Thailand.

Both Bangladesh and Thailand are India’s FTA partners, and trade in yarn and air-

conditioning equipment has been growing rapidly. India’s yarn exports to Bangladesh and

India’s imports of ACE from Thailand have been studied from the point of view of intra-

industry trade (IIT) potential. The computed IIT scores indicate that intra-industry trade

accounted for a moderate share of total trade between India and Bangladesh in the textile

and clothing sector. India’s export of cotton sewing thread (HS 5204), artificial filament yarn

(HS 5403) and synthetic filament yarn (HS 5402) to Bangladesh had relatively high IIT

scores. According to the index scores, the textile and clothing sector offers substantial

Table 12. Panel Granger causality test between trade and LPI

F-Test Null Hypothesis Result

Variables A (X causes Y) B (Y causes X)A (X B (Y Granger

causes Y) causes X) Causality

F-Statistic F-Critical F-Statistic F-Critical F-Critical

Export of yarn 0.759 0.09 0.782 0.08 Reject Reject Bidirectional

and lpi_p

Export of yarn 0.970 0 0.961 0 Reject Reject Bidirectional

and lpi_r

Import of ace 0 62.2 0 52.6 Do Not Do Not No

and lpi_p Reject Reject Causality

Import of ace 0.772 0.08 0.605 0.27 Reject Reject Bidirectional

and lpi_r

66 The usual caveat is that we intentionally ignore running any further panel regression at this point. Ideally,

one may carry a panel regression (e.g. FMOLS), as the variables in questions are co-integrated. Since our

interest is to investigate the causal direction, we concentrated only on Granger causality. Refer Appendix 8 for

a briefed note on Granger causality model.



113

vertical IIT trade opportunities between the two countries. This is most importantly felt in the

Wadding and nonwoven yarns (HS 56), suggesting potential for further intra-industry trade

between the two countries. On the other hand, in the case of India’s import of ACE from

Thailand, IIT index score was relatively high and rising. This is in air, vacuum pumps,

compressors, ventilating fans etc. (HS 8414). However, at present the vertical IIT potentialin ACE does not appear to be very high.

We then explored the likely impact of improved logistical services on a country’s

trade in these two product categories (yarn and ACE), by first developing an index of

logistical performance (LPI). The estimated LPI scores in this study indicate that Asia and

the Pacific region are comprised of a heterogeneous group, which is characterized by wide

gaps in logistics performance. Relatively richer economies occupy the top positions in LPI,

whereas the bottom positions are occupied by the LDCs. For example, Myanmar,

Cambodia, Lao People’s Democratic Republic and Bangladesh occupy the bottom

positions in logistics performance. Other developing countries occupy the middle portion of

the ladder. The logistics gap between the relatively developed and the LDCs in Asia and

the Pacific region appears to have widened more than between 2000 and 2010.

We then proceeded to econometrically estimate the relationship between logistical

performance and trade in the selected products, using both panel regressions and system-

GMM. The result of system-GMM does not reject the static panel data modeling results.

This has enabled us to conclude that improvement in logistic services is associated with

significant increases in trade. We then tested for causality, and found that it runs in both

directions – while better logistics increases trade, more trade tends to also have a positiveimpact on logistical performance.

In terms of policy, this study suggests that efficient performance in logistics

contributes positively to trade in these two products (yarn and ACE), which are important

intermediate goods in manufacturing. This also has the potential to enhance greater

cross-border production linkages, including integrated production networks. Thus, the

improvement in logistics implies that there are mutual gains for countries in the region.

More so, there is room for cooperation to reduce the high logistics gaps, with

implementation of regional logistics sector policy.



114

References

Anderson, J.E., and E. van Wincoop (2004). Trade costs, Journal of Economic Literature,

Vol. XLII, No. 3; pp. 691-751.

Ando, M., and F. Kimura (2009). Fragmentation in East Asia: Further Evidence, ERIADiscussion Paper Series No. 2009-20, ERIA, Jakarta.

Arellano, M., and O. Bover (1995). Another look at the instrumental variable estimation of

error component models, Journal of Econometrics, Vol. 68, No. 1; pp. 29-51.

Arellano, M., and S. Bond (1991). Some tests of specification for panel data: Monte Carlo

evidence and an application to employment equations, Review of Economics

Studies, Vol. 58, No. 2; pp. 277-297.

Arnold, J., (2010). Connectivity and Logistics for the North East and Beyond, in Brunner

(2010), Sage, New Delhi.

Athukorala, P., (2010). Production Networks and Trade Patterns in East Asia:

Regionalization or Globalization?, ADB Working Paper Series on Regional

Economic Integration, No. 56, Asian Development Bank, Manila.

Baldwin, R., (2011). 21st Century Regionalism: Filling the Gap between 21st Century Trade

and 20th Century Trade Rules, Centre for Economic Policy Research Policy Insight

No. 56, London.

Blundell, R., and S. Bond (1998). Initial conditions and moment restrictions in dynamic

panel data models. Journal of Econometrics, Vol. 87, No. 1; pp. 115-143.

Brooks, D.H., (2008). Linking Asia’s Trade, Logistics, and Infrastructure, ADB Institute

Working Paper No. 128, Asian Development Bank Institute (ADBI), Tokyo.

Brooks, D.H., and D. Hummels (eds.). (2009). Infrastructure’s Role in Reducing Asia’s

Trade Costs: Building for Trade, Edward Elgar, Cheltenham.

Brunner, H.P., ed. (2010). North East India: Local Economic Development and Global

Markets, Sage, New Delhi.

De, P., (2009). Empirical estimates of transportation costs: Options for enhancing Asia’strade in Brooks, D.H. and D. Hummels (eds.) Infrastructure’s Role in Lowering

Asia’s Trade Costs: Building for Trade, Edward Elgar, Cheltenham, UK.

De, P., (2011). Why is Trade at Border a Costly Affair in South Asia? An Empirical

Investigation, Contemporary South Asia, Vol. 11, No. 4.

Deardorff, A.V., (2001). International Provision of Trade Services, Trade and Fragmentation,

Policy Research Paper 2548, World Bank, Washington, D.C.

Duval, Y., and C. Utoktham (2011). Trade Facilitation in Asia and the Pacific: Which Policies

and Measures Affect Trade Costs the Most?, Working Paper Series, No. 94, ARTNeT, UNESCAP, Bangkok.

Feenstra, R.C., (2004). Advanced International Trade: Theory and Evidence, Princeton

University Press, Princeton.



115

Findlay, C., (2009). Trade Facilitation in the ASEAN Economic Community, PB-2009-06,

ERIA, Jakarta.

Francois, J., and B. Hoekman. (2010). Services Trade and Policy, Journal of Economic

Literature, September 2010; pp. 642-692.

Granger, C.W.J., (1969). Investigating Causal Relations by Econometric Models and Cross-

spectral Methods, Econometrica, Vol. 37, No. 3; pp. 424-438.

Greenaway, D., R. Hine and C. Milner (1995). Vertical and Horizontal Intra-Industry Trade:

A Cross Industry Analysis for the United Kingdom, The Economic Journal , Vol. 105;

pp. 1505-18.

Hesse, M., and J.P. Rodrigue (2004). The Transport Geography of Freight Distribution and

Logistics, Journal of Transport Geography Vol. 12, No. 3; pp. 171-184.

Hiratsuka, D., and Y. Uchida (eds.) (2008).Vertical Specialization and Economic Integration

in East Asia, IDE-JETRO, Chiba.

Hummels, D., (2007), Transportation Costs and International Trade in the Second Era of

Globalization, Journal of Economic Perspectives, Vol. 21, No. 3; pp. 131-154.

Hollweg, C., and M.H. Wong (2009). Measuring Regulatory Restrictions in Logistics

Services, ERIA Discussion Paper Series, No. 2009-14, Economic Research

Institute for ASEAN and East Asia (ERIA), Jakarta.

Jones, R.W., and H. Kierzkowski (2001). A Framework for Fragmentation, in S. Arndt and

H. Kierzkowski (eds.) Fragmentation, Oxford University Press (OUP), Oxford.

Kimura, F., and A. Mitsuyo (2005). Two-dimensional fragmentation in East Asia: Conceptual

framework and empirics, International Review of Economics & Finance, Vol. 14,

No. 3; pp. 317-348.

Kimura, F., and A. Obashi (2007) International Production Networks in Machinery: Structure

and Its Evolution, ERIA Discussion Paper 2010-09, Economic Research Institute for

ASEAN and East Asia (ERIA), Jakarta.

Kimura, F., and I. Kobayashi (2009). Why is the East Asia Industrial Corridor Needed?,

ERIA Policy Brief No. 2009-01, Economic Research Institute for ASEAN and East Asia (ERIA), Jakarta.

Kimura, F., (2012). Japan’s Mission on Constructing a New International Economic Order,

Japan’s Economic Currents Policy Brief No. 81, Keizai Koho Center, Tokyo.

Kuroiwa, I., (2005). Formation of Inter-country Production Networks in East Asia:

Application of International Input-Output Analysis, paper presented at the 15th

International Input-Output Conference, Beijing, China, 27 June – 1 July.

Mikic, M., and J. Gilbert (2007). Trade Statistics in Policymaking - A Handbook of

Commonly Used Trade Indices and Indicators, United Nations Economic and SocialCommission for Asia and the Pacific (UNESCAP), Bangkok.



116

Planning Commission (2011). Report of the Working Group on Logistics, New Delhi Sims,

C (1972) “Money, Income, and Causality”, The American Economic Review , Vol. 62,

No. 4, pp. 540-552.

Sourdin, P., and R. Pomfret (2012). Trade Facilitation: Defining, Measuring, Explaining, and

Reducing the Cost of International Trade, Edward Elgar, Cheltenham.

Urata, S., and K. Kiyota (2003). Services Trade in East Asia, NBER East Asia Seminar on

Economics (EASE), Volume 11, NBER.

World Bank (2012). Connecting to Compete: Trade Logistics in the Global Economy ,

Washington, D.C.



117

Appendix 1

World Bank LPI, 2012

Country LPI LPI Customs Infras- International Logistics Tracking Time-

Rank Score tructure shipments com- & tracing lines

petence

Australia 18 3.73 3.60 3.83 3.40 3.75 3.79 4.05

Bangladesh * * * * * * * *

Cambodia 101 2.56 2.30 2.20 2.61 2.50 2.77 2.95

China 26 3.52 3.25 3.61 3.46 3.47 3.52 3.80

Hong Kong, 2 4.12 3.97 4.12 4.18 4.08 4.09 4.28

China

Japan 8 3.93 3.72 4.11 3.61 3.97 4.03 4.21

India 46 3.08 2.77 2.87 2.98 3.14 3.09 3.58

Indonesia 59 2.94 2.53 2.54 2.97 2.85 3.12 3.61

Korea, Rep. of 21 3.70 3.42 3.74 3.67 3.65 3.68 4.02

Lao PDR 109 2.50 2.38 2.40 2.40 2.49 2.49 2.82

Malaysia 29 3.49 3.28 3.43 3.40 3.45 3.54 3.86

Mongolia 140 2.25 1.98 2.22 2.13 1.88 2.29 2.99

Myanmar 129 2.37 2.24 2.10 2.47 2.42 2.34 2.59

New Zealand 31 3.42 3.47 3.42 3.27 3.25 3.58 3.55

Pakistan 71 2.83 2.85 2.69 2.86 2.77 2.61 3.14

Philippines 52 3.02 2.62 2.80 2.97 3.14 3.30 3.30

Singapore 1 4.13 4.10 4.15 3.99 4.07 4.07 4.39

Thailand 38 3.18 2.96 3.08 3.21 2.98 3.18 3.63

Viet Nam 53 3.00 2.65 2.68 3.14 2.68 3.16 3.64

Sri Lanka 81 2.75 2.58 2.50 3.00 2.80 2.65 2.90

Note: * Data not available.

Source: The World Bank, Washington, D.C.



118

Appendix 2

HS codes considered calculating the export of yarn to Bangladesh

HS Product Description HS Product Description

code code

5205 Cotton yarn (other than 5603 Nonwovens, whether or not

sewing thread), containing 85% impregnated, coated, covered or

or more by weight of cotton, laminated

not put up for retail sale

5201 Cotton, not carded or combed 5202 Cotton waste (including yarn waste

and garnetted stock)

5208 Woven fabrics of cotton, 5607 Twine, cordage, ropes and cables,

containing 85% or more by weight whether or not plaited or braidedof cotton, weighing not more than and whether or not impregnated

200 g/m2

5509 Yarn (other than sewing thread) 5107 Yarn of combed wool, not put up for

of synthetic staple fibres, not put retail sale

up for retail sale

6006 Other knitted or crocheted fabrics 5508 Sewing thread of manmade staple

fibres, whether or not put up

for retail sale

5407 Woven fabrics of synthetic filament 5007 Woven fabrics of silk or of silk waste

yarn, including woven fabrics

obtained from materials of

heading 54.04

5209 Woven fabrics of cotton, 5404 Synthetic monofilament of 67 decitex

containing 85% or more by weight or more and of which no cross-

of cotton, weighing more than sectional dimension exceeds 1 mm;

200 g/m2 strip and the like (for example,

artificial straw) of synthetic textile

materials of an apparent width notexceeding 5 mm

5402 Synthetic filament yarn (other than 5003 Silk waste (including cocoons and

sewing thread), not put up for unsuitable for reeling, yarn waste

retail sale, including synthetic and garnetted stock)

5510 Yarn (other than sewing thread) 5604 Rubber thread and cord, textile

of artificial staple fibres, not put up covered; textile yarn, and strip and

for retail sale the like of heading 54.04 or 54.05, \

impregnated, coated, covered or

sheathed with rubber or plastics

5504 Artificial staple fibres, not carded, 5002 Raw silk (not thrown)

combed or otherwise processed

for spinning



119

6001 Pile fabrics, including “long pile” 5403 Artificial filament yarn (other than

fabrics and terry fabrics, knitted sewing thread), not put up for retail

or crocheted sale, including artificial monofilament

of less than 67 decitex

5512 Woven fabrics of synthetic staple 5505 Waste, noils, garnetted stock of fibres, containing 85% or more manmade fibers

by weight of synthetic staple fibres

5212 Other woven fabrics of cotton 5606 Gimped yarn, and strip and the like

of heading 54.04 or 54.05, gimped

(other than those of heading 56.05

and gimped horsehair yarn);

chenille yarn (including flock chenille

yarn); loop wale-yarn

5515 Other woven fabrics of synthetic 5601 Wadding of textile materials andstaple fibres articles thereof; textile fibres, not

exceeding 5 mm in length (flock),

textile dust and mill neps

5206 Cotton yarn (other than sewing 5406 Man-made filament yarn (other than

thread), containing less than 85% sewing thread), put up for retail sale

by weight of cotton, not put up

for retail sale

5408 Woven fabrics of artificial filament 5609 Articles of yarn, strip or the like of

yarn, including woven fabrics heading 54.04 or 54.05, twine,obtained from materials of cordage, rope or cables, not

heading 54.05 elsewhere specified or included

5503 Synthetic staple fibres, not carded, 5516 Woven fabrics of artificial staple

combed or otherwise processed fibers

for spinning

5112 Woven fabrics of combed wool or 5608 Knotted netting of twine, cordage

of combed fine animal hair or rope; made up fishing nets and

other made up nets, of textile

materials

5513 Woven fabrics of synthetic staple 5305 Coconut, abaca (Manila hemp or

fibres, containing less than 85% Musa textilis Nee), ramie and other

by weight of such fibres, mixed vegetable textile fibres, not

mainly or solely with cotton, elsewhere specified or included,

of a weight not exceeding 170 g/m2 raw or processed but not spun;

tow, noils and waste of these fibres

(including yarn waste and garnetted

stock)



120

5211 Woven fabrics of cotton, containing 5514 Woven fabrics of synthetic staple

less than 85% by weight of cotton, fibres, containing less than 85%

mixed mainly or solely with by weight of such fibres, mixed

manmade fibres, weighing more mainly or solely with cotton,

than 200 g/m2 of a weight exceeding 170 g/m2

5605 Metallised yarn, whether or not 5602 Felt, whether or not impregnated,

gimped, being textile yarn, or strip coated, covered or laminated

or the like of heading 54.04 or 54.05,

combined with metal in the form

of thread, strip or powder or

covered with metal

5903 Textile fabrics impregnated, coated, 5306 Flax yarn

covered or laminated with plastics,

other than those of heading 59.025210 Woven fabrics of cotton, containing 5005 Yarn spun from silk waste, not put

less than 85% by weight of cotton, up for retail sale

mixed mainly or solely with

manmade fibres, weighing not

more than 200 g/m2

5806 Narrow woven fabrics, other than 5109 Yarn of wool or of fine animal hair

goods of heading 58.07; narrow

fabrics consisting of warp without

weft assembled by means of anadhesive (bolducs)

5401 Sewing thread of manmade 5308 Yarn of other vegetable textile fibres;

filaments, whether or not put up paper yarn

for retail sale

5309 Woven fabrics of flax 5111 Woven fabrics of carded wool or of

carded fine animal hair

5501 Synthetic filament tow 5507 Artificial staple fibres, carded,

combed or otherwise processed for

spinning

5207 Cotton yarn (other than sewing 5103 Waste of wool or of fine or coarse

thread) put up for retail sale animal hair, including yarn waste but

excluding garnetted stock

5204 Cotton sewing thread, whether 5502 Artificial filament tow

or not put up for retail sale

5203 Cotton, carded or combed



121

HS codes considered calculating the import of ACE

from Thailand

HS code Product Description

8415 Air conditioning machines, comprising a motor-driven fan and elements for changing the temperature and humidity, including those machines in which the

humidity cannot be separately regulated

8408 Compression-ignition internal combustion piston engines (diesel or semi-diesel

engines)

8414 Air or vacuum pumps, air or other gas compressors and fans; ventilating or

recycling hoods incorporating a fan, whether or not fitted with filters

BEC codes

BEC Good Description

53 Primary/Semi Processed Transport equipment and parts and accessories

thereof

51 Final Passenger Motor Cars

BEC Good Description

22 Final Processed Industrial Supplies

21 Primary/Semi Processed Primary Industrial Supplies

Production structure of Indian firms in yarn manufacture and

air conditioning equipment

Number of factories Value of Output in Net Value Added in

Commodity Rupees Lakhs Rupees Lakhs

2004 2006 2007 2009 2004 2006 2007 2009 2004 2006 2007 2009

Yarn* 11,342 11,942 11,425 13,417 11,108,327 14,460,645 15,951,148 20,594,695 1,387,909 2,756,030 3,602,741 2,747,245

Air Conditioning

Equipment** 4 ,050 4 ,047 4 ,149 4 ,481 3 ,238,332 5 ,378,400 6 ,836,824 9 ,611,648 671,973 1 ,181,872 1 , 472,233 2 ,437,654

For 2004-2007: NIC-1998 3-digit codes matched with corresponding ISIC Revision 2 codes and HS-1996

4-Digit Codes. *Includes 171 (Spinning, Weaving & Finishing of Textiles), 172 (Manufacture of other

Textiles), 243 (Manufacture of Man-made Fibers); **Includes 291 (Manufacture of General Purpose

Machinery) further disaggregated to 2911 (Manufacture of engines and turbines, except aircraft, vehicle

and cycle engines), 2912 (Manufacture of pumps, compressors, taps and valves), 2919 (Manufacture of

other general purpose machinery). For 2008 onwards: NIC-2008 3-digit codes matched with

corresponding ISIC Revision 2 codes and HS 1996 4-digit codes. *Includes 131 (Spinning, Weaving &Finishing of Textiles), 139 (Manufacture of other Textiles), 203 (Manufacture of Man-made Fibers);

**Includes 281 (Manufacture of General Purpose Machinery) further disaggregated to 2911 (Manufacture

of engines and turbines, except aircraft, vehicle and cycle engines), 2912 (Manufacture of pumps,

compressors, taps and valves), 2919 (Manufacture of other general purpose machinery).



122

Appendix 3

Calculated IIT scores

(Exporter – India, Importer – Bangladesh)

Year Product Product Name IIT

2001 5609 Articles of yarn, strip or the like of heading 54.04 or 0.99

54.05, twine, cordage, rope or cables, not elsewhere

specified or included

2004 5202 Cotton waste (including yarn waste and garnetted stock) 0.93

2000 5601 Wadding of textile materials and articles thereof; 0.91

textile fibres, not exceeding 5 mm in length (flock),


2000 5602 Felt, whether or not impregnated, coated, covered 0.89

or laminated

2007 5608 Knotted netting of twine, cordage or rope; made up 0.89

fishing nets and other made up nets, of textile materials





2003 5204 Cotton sewing thread 0.75



2009 5403 Artificial filament yarn (other than sewing thread), 0.67

not put up for retail sale, including artificial monofilament


2003 5603 Nonwovens, whether or not impregnated, coated, 0.65

covered or laminated



2006 5512 Woven fabrics of synthetic staple fibres, containing 85% 0.65

or more by weight of synthetic staple fibres

2000 5401 Sewing thread of manmade filaments, whether or 0.64

not put up for retail sale

2005 5505 Waste, noils, garnetted stock of manmade fibers 0.59

2001 5103 Waste of wool or of fine or coarse animal hair, 0.58

including yarn waste but excluding garnetted stock





123

2007 5007 Woven fabrics of silk or of silk waste 0.51

2009 5404 Synthetic monofilament of 67 decitex or more and 0.49

of which no cross-sectional dimension exceeds 1 mm;

strip and the like (for example, artificial straw) of

synthetic textile materials of an apparent width notexceeding 5 mm


2006 5606 Gimped yarn, and strip and the like of heading 54.04 0.45

or 54.05, gimped (other than those of heading 56.05

and gimped horsehair yarn); chenille yarn

(including flock chenille yarn); loop wale-yarn




2005 5602 Felt, whether or not impregnated, coated, 0.44

covered or laminated

2005 6001 Pile fabrics, including “long pile” fabrics and terry fabrics, 0.43

knitted or crocheted

2010 5403 Artificial filament yarn (other than sewing thread), 0.42

not put up for retail sale, including artificial monofilament


2007 5210 Woven fabrics of cotton, containing less than 85% 0.41

by weight of cotton, mixed mainly or solely with

manmade fibres, weighing not more than 200 g/m2

2010 5402 Synthetic filament yarn (other than sewing thread), 0.41

not put up for retail sale, including synthetic



2007 5208 Woven fabrics of cotton, containing 85% or more by 0.40weight of cotton, weighing not more than 200 g/m2




2008 5007 Woven fabric of silk or of silk waste 0.37

2002 5407 Woven fabrics of synthetic filament yarn, including 0.36

woven fabrics obtained from materials of heading 54.04.

2003 5607 Twine, cordage, rope and cable 0.34






124




2010 5408 Woven fabrics of artificial filament yarn, including 0.32

woven fabrics obtained from materials of headin

2009 5609 Articles of yarn, strip or the like of heading 54.04 or 0.31

54.05, twine, cordage, rope or cables, not elsewhere

specified or included

2003 5309 Woven fibers of flax 0.30



2006 5406 Manmade filament yarn (other than sewing thread), 0.28

put up for retail sale

2009 5204 Cotton sewing thread, whether or not put up for retail sale 0.27





manmade fibres, weighing more than 200 g/m2

2002 5201 Cotton, not carded or combed 0.22

2008 5602 Felt, whether or not impregnated, coated, covered or 0.20

laminated

2010 5208 Woven fabrics of cotton, containing 85% or more 0.20

by weight of cotton, weighing not more than 200 g/m2



2006 5007 Woven fabric of silk or of silk waste 0.18

2004 5513 Woven fabrics of synthetic staple fibres, containing less 0.18

than 85% by weight of such fibres, mixed mainly or

solely with cotton, of a weight not exceeding 170 g/m2


2007 5513 Woven fabrics of synthetic staple fibres, containing less 0.16

than 85% by weight of such fibres, mixed mainly or

solely with cotton, of a weight not exceeding 170 g/m2



2005 5003 Silk waste (including cocoons and unsuitable for 0.14

reeling, yarn waste and garnetted stock)



125










2003 5408 Knotted netting of twine, cordage or rope; 0.11

made up fishing nets and other made up nets,

of textile materials

2005 5211 Woven fabrics of cotton, containing less than 85% 0.11by weight of cotton, mixed mainly or solely with

manmade fibres, weighing more than 200 g/m2

2009 5806 Narrow woven fabrics, other than goods of heading 0.11

58.07; narrow fabrics consisting of warp without weft

assembled by means of an adhesive (bolducs)




2009 5402 Synthetic filament yarn (other than sewing thread), 0.10

not put up for retail sale, including synthetic



126

Appendix 4

List of logistics performance indicators

Sr. No. Category Indicator Data Source

1 Transport Air transport, freight (million ton-km),

services taken per 1,000 population

2 Air transport, passengers carried,

taken per 1,000 population

3 Container port traffic (TEU: 20 foot

equivalent units), taken per 1,000

population

4 Rail lines (total route-km), taken

per 100 sq km of area

5 Roads, paved, taken as % of total roads

6 Roads, total network (km), taken

per 100 sq km of area

7 ICT services Internet users, taken per 100 population

8 Mobile cellular subscriptions, taken

per 100 population

9 Telephone lines, taken per 100 population10 Financial Domestic credit to private sector,

services taken as % of GDP

11 Human Literacy rate, adult total (% of people

resource quality ages 15 and above)

WorldDevelopment

Indicators (WDI),

World Bank



127

Appendix5

PCAw

eights

Indicators

2000

2001

2002

2003

2004

2005

2006

2007

2008

2009

2010

Airfreighttra

nsport

0.324

0.327

0.322

0.319

0.324

0.324

0.324

0.345

0.333

0

.335

0.353

Airpassenge

rstransport

0.341

0.339

0.343

0.331

0.344

0.348

0.348

0.366

0.349

0

.348

0.331

Containerpo

rttraffic

0.312

0.320

0.323

0.320

0.318

0.315

0.315

0.325

0.310

0

.309

0.318

Raillines

0.044

0.296

0.281

0.272

0.274

0.267

0.270

0.278

0.264

0

.272

0.013

Roads,pave

d

0.281

0.270

0.272

0.277

0.277

0.279

0.280

0.360

0.277

0

.280

0.279

Roads,total

network

0.292

0.252

0.244

0.236

0.241

0.237

0.238

0.243

0.237

0

.228

0.278

Internetuser

s

0.327

0.308

0.317

0.333

0.330

0.326

0.320

0.332

0.321

0

.324

0.346

Mobilecellularsubscriptions

0.380

0.354

0.359

0.361

0.355

0.357

0.359

0.370

0.348

0

.342

0.350

Telephonelines

0.363

0.336

0.338

0.339

0.334

0.337

0.340

0.354

0.341

0

.334

0.334

Domesticcre

dittoprivatesector

0.292

0.284

0.279

0.286

0.281

0.281

0.278

0.275

0.292

0

.294

0.318

Literacyrate

0.219

0.195

0.201

0.205

0.203

0.205

0.208

0.224

0.209

0

.217

0.233

Eigenvalue

6.381

7.139

7.029

6.987

7.046

6.941

6.943

6.512

6.960

6

.831

7.037

Proportionexplained(%)

58.000

64.900

63.900

63.520

64.050

63.10063.120

59.200

63.270

62

.100

63.980



128

Appendix 6

Im, Pesarn, and Shin (IPS) Unit Root Test

Im, Pesaran and Shin (IPS) proposes a test for the presence of unit roots in panels,

and begin by specifying a separate ADF regression for each cross-section with individual

effects and no time trend:

∆yit = αi + ρ

i yi, t 1 + Σ βij ∆yi, t j + ε

it

Where i = 1, . . ., N and t = 1, . . ., T

IPS uses separate unit root tests for the N cross-section units. Their test is based

on the Augmented Dickey-fuller (ADF) statistics averaged across groups. After estimating

the separate ADF regressions, the average of the t -statistics for p1 from the individual ADFregressions, tiTi

( pi):

tNT = ΣtiT (pi βi)

The t -bar is then standardized and it is shown that the standardized t-bar statistic

converges to the standard normal distribution as N and T →∞. IPS (1997) shows that t-bar

test has better performance when N and T are small.

pi

j = 1

1N i = 1

N



129

Appendix 7

Cointegration test of Westerlund

The underlying idea in Westerlund (2007) is to test for the absence of co-integration

by determining whether the individual panel members are error correcting. Consider the

following error-correction model:

D.y_it = c_i + a_i1*D.y_it-1 + a_i2*D.y_it-2 + ... + a_ip*D.y_it-p

+ b_i0*D.x_it + b_i1*D.x_it-1 + ... + b_ip*D.x_it-p

+ a_i (y_it-1 – b_i*x_it-1) + u_it

Where, a_i provides an estimate of the speed of error-correction towards the long run

equilibrium y_it = - (b_i/a_i) * x_it for that series i. The Ga and Gt test statistics test H0: a_i

= 0 for all i versus H1: a_i < 0 for at least one i. These statistics start from a weighted

average of the individually estimated a_i’s and their t-ratio’s, respectively. The Pa and Pt

test statistics pool information over all the cross-sectional units to test H0: a_i = 0 for all i

versus H1: a_i < 0 for all i. Rejection of H0 should therefore be taken as rejection of

co-integration for the panel as a whole.



130

Appendix 8

Granger causality

Testing causality, in the Granger sense, involves using F-tests to test whether

lagged information on a variable Y provides any statistically significant information about

a variable X, in the presence of lagged X. If not, then “Y does not Granger-cause X.” Refer

to Granger (1969), which was popularized by Sims (1972). There are many ways in which

to implement a test of Granger causality. On particularly simple approach uses the

autoregressive specification of a bivariate vector autoregression. Assume a particular

autoregressive lag length p, and estimate the following unrestricted equation by ordinary

least squares (OLS):

x t = c

1 + Σ α i x

t-i + Σ β

i y

t-i + u

t

H D = β 1 = β 2 = ... = β p = 0

Conduct an F-test of the null hypothesis by estimating the following restricted

equation also by OLS:

x t = c

t + Σ y

i x

t-i + e

t

Compare their respective sum of squared residuals.

RSS1 = Σ û2

RSSD = Σ ê2

If the test statistic is greater than the specified critical value, then reject the null

hypothesis that Y does not Granger-cause X.

S1 = ~ F

p, T-2p-1

It is worth noting that with lagged dependent variables, as in Granger-causality

regressions, that the test is valid only asymptotically. An asymptotically equivalent test is

given by:

S1 = ~ x 2 ( p)

Another caveat is that Granger-causality tests are very sensitive to the choice of lag

length and to the methods employed in dealing with any non-stationarity of the time series.

(RSSD – RSS

1)/ p

RSS1 / (T-2p-1)

T (RSSD – RSS1)RSS

1

t = 1

T

t = 1

T

tt

i = 1

p

i = 1

p

i = 1

p



131

III. An analysis of export performance of manufacturingand service sector enterprises in Sri Lanka

By JeevikaWeerahewa, Sarath S. Kodithuwakku and Rifana Buhary

Introduction

Recent theoretical and empirical literature on international trade places renewed

emphasis on the importance of firm level factors as determinants of firms’ export behaviour.

This followed the highly influential Melitz (2003) trade model that highlighted the importance

of firm heterogeneity in determining industry-level trade responses to various shocks.

A large body of empirical evidence indicates that exporting firms are typically larger,

have higher productivity, survive longer, and pay higher wages than non-exporters. Pöschl

et al. (2009) found that size and performance premia (labour productivity and wage) of

exporting firms are significantly higher than those of non-exporters in Austria. Aw et al.

(1999), from a study conducted in Chinese Taipei, revealed that firms with higher

productivity, ex-ante, show a higher tendency to enter into the export market, whereas

exporters with low productivity show a higher tendency to exit from the export market. From

a study of Korean firms, they also found that the differences in productivity do not affect the

entry into or exit from export markets. Salomon and Shaver (2005) revealed that domesticand export sales are interdependent and the factors that influence them are also different.

This was found using estimates of a two-stage least square model for firms in the

manufacturing sector of Spain, covering Spanish-owned firms and foreign-owned firms

during 1990-1997. Salomon and Shaver (2005) discovered that domestic and export sales

complement each other (that is domestic sales positively affect export sales), as Spanish-

owned firms and export sales appear to be driven by pre-existing strengths in the domestic

market. However, for foreign-owned firms, domestic and export sales appear to be

substitutes (that is, domestic sales negatively affect export sales). It appears that foreign-

owned firms, when managing their domestic sales in the larger context of the multinationalnetwork, make trade offs between sales in domestic and foreign markets.

Exporting itself appears to have a positive influence on subsequent firm productivity.

It is argued that this may be due to exporting firms gaining access to technical expertise

from their buyers in terms of new product designs and production methods (Grossman and

Helpman, 1991; World Bank, 1993).

Export performance is influenced by managerial influences (i.e. firm characteristics,

competencies and strategy) and the nature of the external environment (see Cavusgil and

Zou, 1994). Yoshino (2008) provides a good assessment on how domestic supplyconstraints and other firm characteristics explain the geographical orientation of firms’

exports as well as the overall market diversification of African manufacturing exports.

Comprehensive reviews on the determinants of export performance are found in Aaby and



132

Slater (1989), Bilkey (1978), Cheety and Hamilton (1993), Madsen (1987) and Zou and

Stan (1998).

Athukorala and Jayasuriya (1988) and Athukorala et al. (1995) studied the export

performance of Sri Lankan manufacturing firms in mid 1980s and 1990s. However, despite

the major changes in both internal and external environments, as far as the authors’ are

aware, there are no recent studies of the export behaviour of Sri Lankan firms. The

objectives of this study are to address this gap in the literature. Also, the study aims to

(i) characterize exporting and non-exporting firms in the manufacturing and services

sectors of Sri Lanka and (ii) assess the extent to which different firm characteristics and the

external environment within which the firms operate, (i.e. various domestic supply

constraints) explain export performance of the firms (whether the firm sells products directly

or indirectly in the export market).

The rest of the paper is organized as follows. The next section presents the modelused to assess determinants of export performance. In the following section, data used for

the estimation are described. In Section 4 exporting versus non-exporting firms are

characterized. Section 5 presents the results of the estimation, and the final section

presents conclusions and policy implications.

A. A model to assess determinants of export performance

1. Measurement of export performance

Shoham (1996) defines export performance as the result of a firm’s actions in the

export markets. Export performance of a firm is commonly measured by (a) export

propensity, (b) export sales and (c) export intensity. Export propensity is generally defined

as the likelihood of a firm becoming an exporter (Estrin et al., 2008). It is calculated as the

proportion of exporting firms within the total number of firms. Zou and Stan (1998) identified

export sales as the most frequently used measure of export performance. Exporting firms

have two basic options as export channels – direct exports and indirect exports. Exporting

firms often choose indirect exports to minimize transaction costs (Peng and York, 2001).

Hence, the sum of direct and indirect sales in the export market is counted as total export

sales. Export intensity is measured as the share of export sales in total sales (Estrin et al.,

2008; Salomon and Shaver, 2005; Pöschl et al., 2009).

2. Determinants of export performance: An estimation model

The degree of export performance of a firm is determined by both internal and

external factors. Internal determinants of export performance are justified by the resource-

based theory, which conceives a firm as a unique bundle of tangible and intangible

resources (assets, capabilities, processes, managerial attributes, information andknowledge). These unique resources are controlled by the firm, enabling it to conceive and

implement strategies aimed at improving its efficiency and effectiveness (Barney, 1991;

Daft, 1983; Wernefelt, 1984). External determinants of export performance are also justified



133

by the industrial organization theory, which argues that the external factors determine the

firm’s strategy, in turn determines economic performance (Scherer and Ross, 1990). The

logic is that the external environment imposes pressures to which a firm must adapt in

order to survive and prosper (Collis, 1991).

We chose the two-stage Heckman estimation procedure for our econometric

analysis to avoid the potential selection bias, a common problem in studies of export

behaviour of firms. This is because the selected sample firms include those that have

already made the decision to export. The procedure involves a probit model in the first

stage (selection stage) and an OLS in the second stage (outcome stage). The presence of

the selection bias of variables is recognized in the probit model, and correction is done in

the second stage by inserting the calculated correction factor (i.e. inverse Mills ratio) in the

OLS as an instrument.

The Heckman two-step method used in this study is presented below, whereequation 1 is the selection equation and 2 is the outcome equation.

Equation 1: The decision to export (i.e. export propensity) is modeled as a dichotomous

choice.

P ( X i > 0) = 1 if ξ

i > 0;

0 otherwise

ξ i = α + β 1 ∗ BP

i + β 2 ∗ FC

i + β 3 ∗ MC + β 4 ∗ S

i + β 5 ∗ Y + β 6 ∗ Z + β 7 ∗ BC

i + ε

i

Equation 2: The decision of value of export sales or export intensity as an OLS

X i = α + β

1 ∗ BP

i + β

2 ∗ FC

i + β

3 ∗ MC + β

4 ∗ S

i + β

5 ∗ Y + β

6 ∗ Z + β

7 ∗ BC

i + ε

i

The first stage explains the probability that firm i exports, where the dependent

variable is a dummy that is equal to one if a firm exports, zero otherwise. The dependent

variable at the second stage is exports, X i . Two equations were estimated, using two

specifications for X i , the logarithmic value of export value and export intensity of the firm.

BP i is a vector of variables of business performance. The other independent variablesinclude firm characteristics (FC

i ), management characteristics (MC ), size dummies (S

i ),

industry dummies (Y ), provincial dummies (Z ) and behind the border constraint dummies

(BC i ). ε

iis an error term assumed to be independently and identically distributed.

B. Data description

1. Enterprise survey of the World Bank: Coverage

The World Bank’s Enterprise Surveys provide a unique source of information that

can be used to analyse the degree of export performance and its determinants. The most

recent survey was conducted in 2011 and it covered 836 enterprises in the manufacturing

and service sectors from all the nine provinces of Sri Lanka. The sample was selected



134

using stratified random sampling. Three levels of stratification were used: industry, size,

and province. (See Appendix 1 for details on sampling method.)

2. Data used for measurement of export performance

Export intensity was measured as the proportion of exports (both direct and indirect)in the firm’s total sales. The value of exports was measured by multiplying the export

intensity by the sales value. Export market share of the firm is the percentage of export

sales in total exports. The basic measure of labour productivity used is value-added per

worker. This is measured as the total sales of the firm less the cost of the raw materials and

intermediate inputs used to produce the output, divided by the number of production

workers in the firm.

3. Data used for the estimation of the model

The firm-specific characteristics include variables such as age of the firm, legal

status and possession of internationally recognized quality certification. The number of

years of experience is included as a management characteristic. Size of the firm is

measured using number of employees. Industry dummies are defined using standard

classifications in the Enterprise Surveys. Provincial dummies are included to take into

account geographical variations in exports. “Behind the Border constraints” (included as

dummy variables) included the managers’ perceptions about the major problems impacting

on the business environment with respect to customs and regulations, business licensing

and permits, access to finance, corruption, courts, crime, electricity, inadequately educated

workforce, labour regulations, informal sector competition, political instability, tax

administration and tax rates. These were assumed to be beyond the control of the

managers, while managerial perceptions about problems that can be addressed are

hypothesized to enhance export performance.

C. Characterization of exporting firms versus non exporting firms

1. Export propensity of firms

Figure 1(a) shows that a higher percentage of non-exporters are found among small

and medium firms, while large firms include nearly equal numbers of exporters and non-

exporters. Figure 1(b) clearly shows that exporters are concentrated in firms located in

the Southern, Central and Western provinces. Industry-wise categorization depicted in

figure 1(c) indicates that food industry has a comparatively higher number of exporters

compared to other industries.

Figures 2(a)-2(c) show the kernel density functions of total sales, labour

productivities and wages, in log terms, for exporters and non-exporters, respectively.

Though total sales, labour productivities and wages of both exporters and non-exportersappear to be distributed normally, the curves of the exporters are shifted to the right

compared to those of non-exporters. The average labour productivity of exporters is on

average 46 per cent higher than that of non-exporters, and exporting firms pay 45 per cent

higher wages.



135

In_wage

-3

-2

-1

0

10 15 20 25

Exporters Non-exporters


100

90

80

70

60

50

40

30

20

10

0

P e r c e n t

S m a l l

M e d i u m

L a r g e

A l l

Size Province Industry

C e n t r a l

E a s t e r n

N o r t h - C e n t r a l

N o r t h e r n

N o r t h - W e s t

S a b a r a g a m u w a

S o u t h e r n

U v a

W e s t e r n A

l l

F o o d

I C T

M a n u f a c t .

O t h e r S e r v .

A l l

100

9080

7060

50

40

3020

10

0

P e r c e n t

100

9080

70

60

50

40

30

20

10

0

P e r c e n t

Exporters Non-exporters Exporters Non-exporters

Figure 1(a). Exporters vs

non-exporters by size

Figure 1(c). Exporters vs

non-exporters by industry

Figure 1(b). Exporters vs

non-exporters by province

Figure 2(a). Kernal density

of log sale

Figure 2(b). Kernal density of log labour

productivity

Figure 2(c). Kernal density of log wage


10 15 20 25

-25

-2

-15

-1

-05

0

In sale

.4

.3

.2

.1

.010 12 14 16 18 20

In labour productivity




136

2. Characteristics of exporting firms

Exporting firms constitute only 10.65 per cent of the sample, indicating that only

a small proportion of firms in Sri Lanka are exporters. Of the exporters, 40 per cent are

large firms, 44 per cent are medium and 16 per cent are small firms (table 1) and the

proportion of export firms in total firms of large, medium and small firms export are 30 per

cent, 14 per cent and 3 per cent respectively. Table 2 provides details of firm heterogeneity

by industry. Of the exporting firms, 36 per cent are in manufacturing, 34 per cent are in

food, 24 per cent are in information communication technology (ICT) related industries and

7 per cent are in other service industries.

Table 1. Heterogeneity of firms by size

Size Number of Share of Number of Export Share of

Firms Firms Export propensity Export Firms

Firms (Share of out of Total

Export Firms Export Firms

out of Total

Firms)

Large 119 14.23 36 30.25 40.45

Medium 278 33.25 39 14.03 43.82

Small 439 52.51 14 3.19 15.73

Total 836 100.00 89 10.65 100.00

Small > = 5 and < = 19 workers; Medium > = 20 and < = 99; Large > = 100 workers

Table 2. Heterogeneity of firms by industry

Industry Number of Share of Number of Export Share of

Firms Firms Export propensity Export Firms

Firms (Share of out of Total

Export Firms Export Firms

out of Total

Firms)

Food 121 14.47 30 24.79 33.71

Health 124 14.83 0

ICT 120 14.35 21 17.50 23.60

Manufacturing 231 27.63 32 13.85 35.96

Other services 116 13.88 6 5.17 6.74

Tourism 124 14.83 0

Total 836 100.00 89 10.65 100.00



137

Table 3 presents the results of the analysis of export performance of various

industries located in various geographical locations. Food industry firms, of which 41 per

cent are exporters, are mostly located in Southern and Central provinces. ICT and other

service industries are mainly concentrated in the Western province and they are mostly

domestic market oriented.

There is considerable variation in the proportion of export sales in total sales, i.e.

export intensity, among firms by location and industry (table 3). Textile exporting firms in the

Western, Central, Sabaragamuwa and North-West provinces export 100 per cent of their

total production. Food exporting firms in all provinces export nearly all their total production,

with the exception of firms in the North-West province. Firms that produce and export

fabricated metal products in the Central province have low export intensities. In contrast,

firms in the Southern province exporting non-metallic mineral products only target the

foreign market.

Table 3. Export propensity, intensity by industry and province

Industry Sub-category Province No. of Export Export

Export propensity intensity

Firms (%) (%)

Food Food Southern 11 40.74 97.61

Central 7 41.18 98.24

North-West 2 15.38 82.29Sabaragamuwa 5 29.41 100.00

Uva 1 100.00 100.00

Western 4 16.67 99.90

Light Fabricated metal Western 3 8.33 79.46

manufacturing products Central 1 8.33 1.00

Machinery and Western 4 14.29 20.67

equipment

Other Basic metals Central 1 100.00 95.00manufacturing Non-metallic Southern 3 75.00 100.00

mineral products North-West 3 25.00 64.40

Paper Western 1 50.00 25.00

Plastics and rubber Southern 1 100.00 25.00

Central 1 100.00 7.00

Textiles Southern 3 37.50 91.63

Western 3 50.00 100.00

Central 3 75.00 100.00

Sabaragamuwa 2 33.33 100.00

North West 1 100.00 100.00



138

Export ConcentrationPer cent of exports

100

90

80

70

60

50

40

30

20

10

0

0 10 20 30 40 50 60 70 80 90 100

Per cent of Firms

Ranked by total sales Ranked by export sales

The data on export performance by ownership and size class revealed that of 43

firms with some foreign ownership, 10 are exporters and they account for 27 per cent of

total exports. Interestingly, the top 10 per cent of the exporters are responsible for more

than 80 per cent of aggregate exports. In other words, a few firms – “superstar exporters” –

are responsible for the bulk of exports.

The concentration of exports across export oriented firms is shown in Figure 3 usingLorenz curves. The figure depicts that the contribution by small firms to the total export

value is insignificant. The two curves, i.e. total sales and total exports almost coincide,

indicating the strong correlation of exports and overall size of the firms.

Tobacco Sabaragamuwa 1 33.33 80.00

Wood Western 1 33.33 25.00

ICT IT Western 15 21.43 33.41

Electronics Western 6 16.22 10.60

Other services Wholesale Western 1 20.00 79.00

Retail Western 1 11.11 25.00

Central 1 12.50 100.00

Transport Western 2 66.67 91.67

Construction Western 1 50.00 10.00


Industry Sub-category Province No. of Export Export

Export propensity intensity

Firms (%) (%)

Figure 3. Export concentration by value of exports and total sales



139

D. Results of the estimation of Heckman Model

Descriptive statistics of the variables used in the two-stage Heckman selection

model are reported in table 4.

Table 4. Descriptive statistics of variables used in the Heckman estimation

Variable Category Description Obs. Mean SD Range

Sales Total annual sales 647 323 2 130 0.4-

(Rs. Million) 35 000

Export Total annual export 89 504 1 920 0.05-

(Rs. Million) 16 000

Export intensity Export intensity 89 65.45 38.52 1-100

(per cent) Age Firm age (years) 820 20.14 17.22 0-167

Manager experience Years of experience 695 18.10 10.94 1-50

(years)

Size Large 0 if the firm size is small 836 0-1

Medium

Province Western 836 0-1

Central

Southern

Industry Food 836 0-1

ICT

Manufacturing

Legal Company 836 0-1

status Partnership

Biggest Custom 683 0-1

obstacle License

Electricity

Access to

finance

Political

instability

Tax

administration

Quality 1 if the firm has 694 0-1

certification quality

certification

0 if the province is

Eastern, North-Central,

North-West, Northern,Sabaragamuwa

and Uva

0 if the industry is health,

tourism and other

services

0 if the legal status is

sole proprietorship and

other

0 if the biggest obstacle

is corruption, courts,

crime, inadequately

educated workforce,

labour regulations,

informal sector

competition and tax rate



140

The estimation results are summarized in table 5. As indicated earlier, two sets of

models are estimated with different specifications for the dependent variable in the

outcome equation, while the selection equation is the same in both models. The same set

of independent variables is used in both selection and outcome equations in the two

models.

In general, the results of the estimation of the first stage of the Heckman Model

(selection model) indicate that (a) large and medium firms (in terms of total sales and

number of workers) have a higher export propensity, (b) food, manufacturing and ICT

industries have a higher tendency to become exporters, (c) firms in Southern and Central

provinces predominate in exporting, (d) the legal status as a company and partnership are

associated with the decision to export, and (e) the young firms have a higher likelihood of

being exporters (i.e. new firms tend to be more likely to export than older firms).

The estimation results of the second stage of the Heckman Model indicate that thevalue of exports is significantly influenced by the size of the firm, as measured by sales and

type of the industry. Furthermore, results show that a 1 per cent increase in total sales is

associated with more than 0.75 per cent increase in exports. According to model 1, the

value of exports of firms which possess an internationally recognized quality certification

are higher.

Managers’ perceptions of the external business environment, such as their views

about the degree of political instability, difficulty in obtaining business licences and permits

etc., have significant negative effects on the value of exports and export intensity. However,the results indicate such perceptions do not have a statistically significant effect on export

market participation.

There is no statistically significant difference on export participation between the

firms that perceived access to finance as a major obstacle and the other firms. However, it

is very interesting that those firms who decided to engage in exporting despite seeing

access to finance as a barrier, demonstrated better performance, in terms of both value and

intensity of exports. This may be due to such firms being more proactive in searching for

and capitalizing upon export growth opportunities that would eventually ease their financing

constraints, however we cannot investigate this issue further with the available data.



141

Table 5. Results of the Heckman Model estimation

Variable Category Export Log of Mills Export Mills

Participation Value of Model 1 intensity: Model 2

Export: Model 2

Model 1

Log (Sales) 0.128** 0.751*** -6.613*

(0.058) (-0.122) (3.463)

Size Large 0.773** 0.652 31.481*

(0.322) (0.613) (17.449)

Medium 0.534** 0.780 26.477*

(0.226) (0.509) (14.600)

Industry Food 1.492*** -0.713 0.424

(0.293) (1.119) (32.099)ICT 1.122*** -2.145** -51.414*

(0.316) (1.031) (29.582)

Manufacturing 1.340*** -0.933 -16.833

(0.271) (1.068) (30.664)

Biggest Custom 0.155 -0.392 -9.583

obstacle regulations (0.380) (0.488) (13.797)

Electricity 0.432 -0.429 -1.100

(0.293) (0.424) (11.974)

Access to -0.022 0.727* 20.061*

finance (0.249) (0.402) (11.499)

License 0.161 -1.826*** -39.593**

(0.398) (0.603) (17.268)

Political 0.354 -2.515*** -65.725***

instability (0.581) (0.784) (22.161)

Tax 0.185 -0.577 -10.822

administration (0.299) (0.406) (11.483)

Province Central 0.936*** -1.029 -10.737(0.349) (0.764) (21.859)

Sabaragamuwa 0.510 -0.166 4.752

(0.380) (0.687) (19.704)

Southern 0.963*** -0.232 1.344

(0.319) (0.755) (21.572)

Western 0.411 -0.534 -16.508

(0.306) (0.611) (17.601)

Legal status Company 0.569** -0.029 1.228(0.232) (0.493) (14.086)

Partnership 0.421* -0.475 -14.745

(0.235) (0.451) (12.891)



142

Conclusions and policy implications

This analysis of the export behaviour of firms in Sri Lanka revealed significant

differences among exporting and non-exporting firms, as well as among exporting firms

themselves. These are likely to generate significant disparities. The Sri Lankan export

sector is clearly dominated by a few large firms, who have larger total sales, pay higher

wages and have higher labour productivity. The study also found that managers’

perceptions of problems in the business environment, associated with ‘behind the border’

problems, negatively impact on export performance. These highlight the fact that Sri Lanka

must improve its domestic business environment and create a more conducive

environment for SME’s to become involved in export markets.

Log -0.245**

(Firm age) (0.119)

Log 0.166 0.310 8.584

(Manager (0.128) (0.192) (5.452)

experience)

Quality 0.241 0.603* 10.970

certification (0.233) (0.325) (9.175)

Lambda -0.518 -6.906(0.894) (25.479)

Constant -5.405*** 4.415 173.680

(1.043) (4.920) (140.570)

Observations 585 585 585 585 585

Standard errors in parentheses and *** p < 0.01, ** p < 0.05, * p < 0.1


Variable Category Export Log of Mills Export Mills

Participation Value of Model 1 intensity: Model 2

Export: Model 2

Model 1



143

References

Athukorala, P., S. Jayasuriya and E. Oczkowsk (1995). Multinational Firms and Export

Performance in Developing Countries: Some Analytical Issues and New Empirical

Evidence. The Journal of Development Economics. 46: 109-122. Athukorala, P., and S. Jayasuriya (1988). Parentage and Factor Proportions: A Comparative

Study of Third-World Multi Nationals in Sri Lankan Manufacturing, Oxford Bulletin of

Economics and Statistics, Vol. 50, No. 4, November 1988: 409-423.

Aw, B.Y., S. Chung, and M.J. Roberts (1999). Productivity and Turnover in the Export

Market: Micro-level Evidence from the Republic of Korea and Taiwan (China), The

World Bank Economic Review , Vol. 14 No. 1; pp. 65-90.

Aaby, N.E., and S.F. Slater (1989). Management Influences on Export Performance: A

Review of the Empirical Literature 1978-88, International Marketing Review , Vol. 6No. 4; pp. 7-26.

Barney, J., (1991). Firm Resources and Sustained Competitive Advantage, Journal of

Management. Vol. 17; pp. 99-120.

Bilkey, W.J., (1978). An Attempted Integration of the Literature on the Export Behavior of

Firms, Journal of International Business Studies, Vol. 9, Spring/Summer; pp. 33-46.

Cavusgil, S.T. S., and Zou. (1994). Marketing Strategy-Performance Relationship: An

Investigation of the Empirical Link in Export Market Ventures, Journal of Marketing ,

Vol. 58; pp. 1-21.

Chetty, S.K., and R.T. Hamilton (1993). Firm-level Determinants of Export Performance:

A Meta-analysis, International Marketing Review , Vol. 10 No. 3; pp. 26-34.

Collis, D.J., (1991). A Resource-based Analysis of Global Competition: The Case of the

Bearings. Industry, Strategic Management Journal , Vol. 12; pp. 49-68.

Daft, R., (1983). Organization Theory and Design, West Publishing, New York.

Estrin, S., K.E. Meyer, M. Wright, and F. Foliano, F. (2008). Export Propensity and Intensity

of Subsidiaries in Emerging Economies, International Business Review , Vol. 17;

pp. 574-586.

Grossman G., and E. Helpman (1991). Innovation and Growth in the Global Economy, MIT

Press, Cambridge MA.

Madsen, T.K., (1987). Empirical Export Performance Studies: A Review of

Conceptualizations and Findings, in Cavusgil, S.T. and Axinn, C. (Eds.), Advances

in International Marketing, JAI Press, Greenwich, CT, Vol. 2; pp. 177-98.

Melitz, M., (2003). The Impact of Trade in Intra-industry Reallocations and Aggregate

Industry Productivity. Econometrica 71(6), 1695-1725.

Peng, M.W., and A.S. York (2001). Behind Intermediary Performance in Export Trade:

Transactions, Agents, and Resources, Journal of International Business Studies,

Vol. 32, No. 2; pp. 327-346.



144

Pöschl, J., R. Stehrer and R. Stöllinge (2009). Characteristics of Exporting and Non-

exporting Firms in Austria, FIW Research Reports 2009/10.

Salomon, R., and J.M. Shaver (2005). Export and Domestic Sales: Their Interrelationship

and Determinants. Strategic Management Journal , Vol. 26; pp. 855-871.

Scherer, F.M., and D. Ross (1990). Industrial Market Structure and Economic Performance,

Rand McNally, Chicago, IL.

Shoham, A., (1996). Marketing-mix Standardization: Determinants of Export Performance,

Journal of Global Marketing , Vol. 10 No. 2; pp.3-73.

Wernerfelt, B., (1984). A Resource-based View of the Firm, Strategic Management Journal ,

Vol. 5, No. 2; pp. 171-80.

World Bank (1993). The East Asian Miracle: Economic Growth and Public Policy, Oxford:

Oxford University Press for the World Bank.

Yoshino, Y., (2008). Domestic Constraints, Firm Characteristics and Geographical

Diversification of Firm-Level Manufacturing Exports in Africa. Policy Research

Working Paper 4575, World Bank.

Zou, S., and Stan, S. (1998). The Determinants of Export Performance: A Review of the

Empirical Literature between 1987 and 1997. International Marketing Review ,

Vol. 15, No. 5; pp. 333-356.



145

Appendix 1

Method of sampling used in the enterprise survey

The sample for Sri Lanka is selected using stratified random sampling. Three

levels of stratification are used in this country: industry, establishment size, and region.

Industry stratification is designed as follows. The universe is stratified into 2

manufacturing industries, 1 services industry (retail), and 2 residual sectors as defined in

the sampling manual. Each manufacturing industry had a target of 120 interviews. The

services industry and the 2 residual sectors had a target of 120 interviews.

Size stratification is defined following the standardized definition for the rollout;

small (5 to 19 employees), medium (20 to 99 employees), and large (more than 99

employees). For stratification purposes, the number of employees is defined on thebasis of reported permanent full-time workers. This seems to be an appropriate

definition of the labour force since seasonal/casual/part-time employment is not

a common practice, except in the sectors of construction and agriculture.

Regional stratification is defined in nine regions: Eastern, Western, Southern,

Central, Northern, North-Central, North-West, Uva, Sabaragamuwa.

One frame is used for the Enterprise Survey in Sri Lanka. The sample frame

containing fresh contacts used in the Sri Lanka is obtained from the Department of

Census and Statistics of Sri Lanka (DCS) 2003.

The enumerated establishments are then used as the frame for the selection of

a sample with the aim of obtaining interviews at 600 establishments with five or more

employees.

The quality of the frame is assessed at the onset of the project through calls to

a random subset of firms and local contractor knowledge. The sample frame is not

immune from the typical problems found in establishment surveys: positive rates of

non-eligibility, repetition, non-existent units, etc. Due to response rate and ineligibilityissues, additional sample had to be extracted by DCS and the World Bank in order to

obtain enough eligible contacts and meet the sample targets.

Given the impact that non-eligible units included in the sample universe may

have on the results, adjustments may be needed when computing the appropriate

weights for individual observations. The percentage of confirmed non-eligible units as

a proportion of the total number of sampled establishments contacted for the survey is

50 per cent (900 out of 1806 establishments).

Source: World Bank



147

IV. Evaluation of business association membershipon small and medium enterprises’ growth performance:

Evidence from enterprise survey of Cambodia

By Vathana Roth

Introduction

There is growing attention on small and medium enterprises (SMEs) within

academic circles and policy discussions. It is widely argued that SMEs play a vital role in

enhancing economic growth and competitiveness, helping to reduce poverty in both

developing and industrialized economies. The effects are particularly profound indeveloping countries, where SMEs contribute a large portion of domestic production and

employment (Ayyagari et al., 2011).

However, SMEs face a number of constraints to growth. This has to do with their

small size, relatively low bargaining power, difficulty obtaining investment loans,

vulnerability to idiosyncratic risks (resultant of vertical and horizontal competing forces),

high costs of administrative compliance and weak legal enforcement (e.g. Sukiassyan and

Nugent, 2008; Beck et al., 2008; Stephanou and Rodriguez, 2008; Aterido et al., 2009). To

cope with these problems, SMEs need specific direct and indirect assistance to improvesurvival rate by becoming more competitive. Under a SME-tailored policy, there are support

programmes, including training, advice, subsidies, for SME owner-managers to take

advantage of (Wren & Storey, 2002; Batra and Mahmood, 2003; Bennett, 2008; Zecchini

and Ventura, 2009; Han and Benson, 2010; Czarnitzki and Hottenrott, 2011; World Bank,

2010; Chheang et al., 2011). Along with programmes designed to fit the needs of individual

SMEs, there are more concentrated and associative programmes to address common

constraints. This has been implemented to mobilise collective efforts and pooled resources.

Examples of associative networks include business and trade associations, professional

and employer associations, federations, networks and clusters.

There are a number of associative organisations, which vary greatly in dynamism

and diversity. Spanning regional, national, local and sectoral levels, the enterprise

networks’ work is through advocacy and representation of members’ interests to external

parties. The aim is to help members be more competitive and be involved in overall

business activities. There is substantial literature that argues that business associations

(BAs) provide more benefits to SMEs rather than to large firms. A larger firm’s economy of

scale and scope enables them to overcome size constraints and stay competitive even

without assistance (Bennett, 1998; Bennett and Ramsden, 2007; Perry, 2007).However,

there are theoretical and empirical challenges to this claim.



148

In Cambodia, SMEs have contributed significantly to overall macroeconomic

growth, job creation, and, to a lesser extent, innovation and technology adoption. The

policy agenda and support programmes are similar to those commonly observed in other

economies. SMEs, particularly manufacturing establishments, are important to economic

growth and business competitiveness. Cambodian manufacturing SMEs have graduallygrown in number, with 33,195 manufacturing establishments in 2007, making up 98.3 per

cent of total number of SMEs. Large enterprises (LEs) account for just 1.7 per cent (NIS

2008). Preliminary results of the 2011 economic census reports that MSMEs account for

502,372 (99.9 per cent) of the total 503,008 establishments across the country. Most of

these firms are concentrated in Phnom Penh and in major economically active provinces,

such as Kampong Cham, Siem Reap, Battambang, Kandal and Takeo (NIS, 2011).

The number of SMEs has risen at an annual rate of about 3.5 per cent since 1998,

reaching 32,619 establishments in 2007 (NIS, 2008). Of the total establishments in 2007,

SMEs involved in food, beverages and tobacco made up the largest proportion (80.9 per

cent). This is followed by those dealing in fabricated metal products and textiles (8.8 per

cent), wearing apparel and leather goods (4.5 per cent). In the same year, the number of

SME employees increased by 9.0 per cent compared to a year earlier, from 87,072 workers

to 94,835. This represents an 18.9 per cent share of total employment (NIS, 2008). The

total volume of production generated by SMEs also constantly climbed at an annual growth

rate of 30.2 per cent between 1998 and 2007, reaching $636.2 million in 2007. Food,

beverages, and tobacco continued to comprise the biggest proportion of total generated

value (NIS, 2008).

Cambodian SMEs, however, are still in initial stages of development. This is despite

the completion of the first five-year SME Development Framework (2005-2010). Initiatives

in the previous plan were designed to address issues and challenges SMEs face. However,

rigorous implementation was lacking and unsatisfactory, and progress was slow.

Additionally, Cambodia lacks systematically study and evaluation on whether the proposed

SME policies, and other related interventions, are having positive impacts on SMEs’

operation and competitiveness.

There are a number of public sponsored or privately financed Business Associations (BAs), as well as growing effort from both government and development

partners to enhance inter-firm collaboration. But there are no systematic studies on whether

this has, or will have, positive impact on participating firms. Therefore, this study aims to

examine the potential impacts that SME membership in BAs or CoC would have on a firm’s

medium and long-term outcomes. The study contributes to the limited, but growing, body of

literature67 on SME development in Cambodia. It also serves stakeholders decision-making

by utilising evidence-based impact evaluation.

67 Some of the qualitative and quantitative studies on Cambodian SMEs are: Harner, 2003; Meas, 2006;

Baily, 2007; Shariff and Chea, 2008; Harvie et al., 2010; Chheang et al., 2011.



149

The rest of the paper is organised as follows. Section 2 reviews and discusses

academic studies on the roles of BAs and the benefits they might provide to participating

firms. It also highlights theoretical and analytical bases for empirical analysis. Section 3

specifies econometric models and variables measurement to estimate ATE of BA

membership on intermediate outcomes and long-term growth performance of SMEs andfirms in general. Section 4 provides data source and a brief descriptive analysis. Section 5

discusses main findings and highlights sensitivity analysis. Section 6 outlines study’s

limitations, and Section 7 concludes the paper and recommends areas for further research.

A. Literature review and theoretical aspects

Moore and Hamalai (1993) defines BAs as a collection of formal membership

organizations of individual business people or firms. Managerial modalities of BAs are

diverse and encompass a huge variation of strategic services and activities. BAs can be bigor small; local, provincial, national, regional or global. They can encompass all firms or just

be sector-specific; publicly or privately owned; independent, government-sponsored; have

voluntary or obligatory memberships; are of discriminatory membership or are open to all

establishments. BAs can offer professional services tailored to the needs of individual

members, which demand more investment in human resources, or provide more services

for collective purposes, which may give room for members to free-ride (e.g., Moore and

Hamalai, 1993).

Potential services of BAs include (but are not limited to) dissemination of marketinformation, government lobby, information on import and export markets, access to

finance, updates on laws and regulations, collective procurement and sales. Services could

also include experience sharing in management practices, production methods, financial

reporting, and short-term vocational and technical training. The majority of associations

undertake advocacy work. The availability of various services to members is as significant

as the effectiveness of service delivery. This is due to members having higher demands on

services deemed more essential.

Two theoretical perspectives can be used to examine and explain the applicability of

BAs or other kinds of collective bodies. Its existence is believed to either contribute to

members’ growth performance or jeopardise socially optimal benefits. These two theory are

pluralism and the public choice (see, for example, Becker, 1985; Moore and Hamalai, 1993;

Goldsmith, 2000: 40-41). The pluralist theory of interest groups gives importance to BAs,

and other formally organized groups, where members can collectively increase political and

economic bargaining power and influence public policy to improve the overall business

environment. Public choice theorists see such associations as counter-productive and

discriminatory due to their rent-seeking behaviour which benefits special interest groups at

the expense of the majority.

The pluralists argue that in order to create a business climate conducive to firms’

growth, government-private sector collaboration is inevitable. This is to reduce transaction

costs of acquiring information on updates and progress of public policies, as well as to



150

ensure a more symmetric flow of information on rules and regulations. Furthermore,

pluralists point to the inability and ineffectiveness of individual firms in influencing

government policy. This claim is much more relevant to micro firms and those SMEs whose

small size may not give them sufficient voice when they are acting alone. According to

pluralists, another benefit of BAs or associative organizations is the expansion and depth of social capital that can be achieved through inter-firm collaboration and networks.

On the other hand, public choice theorists criticize special interest groups and

lobbyists, citing their rent-seeking behaviour as having potential to generate socially-not-

optimal results. Organized special interest groups can benefit from additional economic

rents68 resulting from policy lobby to favour their specific agendas. Thus, in the view of

public choice theorists, BAs and other collective organizations obstruct fair and just

business competition. They say that the representative role of BAs draws government to

fulfil certain business needs, which are more likely to favour members than society at large.

The rent-seeking behaviour may also leave room for bribery and corruption.

In this respect, cautious must be observed when allowing the establishment of

associations or special interest groups. However, the enormous number of BAs and other

associative organizations – labour unions, industrial clusters, and federations – in both

developing and developed economies supports the pluralists’ claim that such organizations

tend to provide more advantages to members than disadvantages. Even if BAs might

generate negative effects under specific circumstances, the impact on overall business

activity might not be completely socially disruptive. Also, BAs members have a wide variety

of strategic services and activities at their disposal, not just policy lobbying.

Empirically, the benefits of BA membership have been well documented.

Considerable literature argues that intended services of BAs are necessary to help member

firms resolve collective or individual problems (Levitsky, 1992; Bennett, 1995, 1998; Nadvi,

1999; Doner and Schneider, 2000, 1999; Luna and Tirado, 2008; Goldsmith, 2000). Political

and economic benefits of BAs have also been documented in the United States, where BAs

are comparatively small, unorganized and fragmented (McCormick et al ., 2008). Benefits of

BAs are even more profound for the smallest firms and SMEs (Wilts and Meyer, 2005).

Levitsky (1992) proposes that the roles of BAs should be enhanced in developing

countries. More so, government, private sector and international organizations should

provide further assistance for BAs. Doner and Schneider (1999) argues that, rather than

impeding market competition and economic growth resulting from collective bargaining

power, BAs could provide a wide range of services. This could be from lowering information

and transaction costs to upgrading skills and technology of member firms. The collective

actions of BAs can be more beneficial for SMEs than for large firms, who can exploit

economies of scale (Wilts and Meyer, 2005; Bennett, 1998).

68 Goldsmith (2000) defined economic rents as “policy-induced gains that would not exist in a competitive

market.”



151

However, there is continuing debate about BAs in terms of service quality, size,

efficacy and effectiveness, institutional arrangements, forms of and motives for membership

(Bennett and Ramsden 2007). The theory of associative organizations has differentiated

two forces that have the potential to drive association membership – “the logic of collective

influence” and “the logic of services” (Olson, 1965; Bennett, 1998, 2000; Perry, 2007). Theformer stresses collective activity in favour of all, or the majority of, members with third

parties (such as government agencies) that can impact on members’ interests. The latter

emphasises service provision to meet specific needs where a BA’s secretariat responds to

members’ requests and enquiries. This allows the BA to assist members in raising and

enhancing competitiveness in “niche” markets in which members specialise (Bennett,

1998).

The distinctive features of institutional arrangement of BAs are resources, size,

effectiveness, service coverage, quality, membership and membership fees. These

differences have implications on the scope and coverage of BAs’ services, social, political

and economic powers and resources. Levitsky (1992) claims that publicly financed and

centralised institutions (staffed and managed by “government appointees”) are less

effective in providing intended services and training to SMEs. He points out that these

institutions often lack human resource capacity and require budgetary resources to properly

fulfil their mandates. He proposes that in developing countries, attention and examination

should be given to those private sector organizations that can more effectively address the

needs of SMEs.

Membership in most CoCs or BAs are often voluntary, however this varies acrosscountries. Germany, France, Austria, the Netherlands and Italy, for instance, require

businesses to register with the CoC (Bennett 1995). In Germany, the Chamber of Industry

and Commerce and the Chamber of Crafts represent business operators in specific local

areas, providing a wide range of services from technical services to government lobbying

(Germany Trade and Invest, 2012: 34). These associative organizations are supported by

public law status and are publicly financed. Japan uses a mixed structure of voluntary and

compulsory membership. However, given strong social ties and pressure, almost all SMEs

(95 per cent) are members of associations (Levitsky, 1992). Most associative organizations

in the US and UK are privately operated with voluntary memberships. Voluntary or statutorymembership could have implications for BAs’ service provision and financing sources.

Bennett (1995) shows that, given voluntary membership, the UK CoC tends to provide

specific services demanded by members, rather than collective services that other

members might use. The logic of services is more relevant when membership in BAs is

voluntary. This is to guard against free-riding among both members and non-members.

However, BAs might face financial constraints as a result of this voluntary membership, as

they mainly rely on membership fees.

It is clear that the issues facing the existence and operation of BAs are multifaceted.These issues need to be continually addressed to create a business environment that is

more inclusive and fostering fair competitive opportunities for all. As an initial step, this

study will attempt to quantify the benefits of private BA membership on firms’ outcome



152

indicators in the Cambodian context. The institutional and financial arrangements of BAs

are to be left for subsequent studies.

B. Econometric specifications and variables measurement

1. Econometric specifications

This study adopts a non-experimental approach, using a variety of econometric

methods to measure average treatment effects of membership in BAs or CoCs on

a number of firms’ intermediate and long-term outcomes. This includes propensity score

matching (PSM) and PSM with ordinary least squares (OLS) regression. The purpose of

using different, but interrelated, approaches to estimation is in order to address

inconsistency and bias of coefficients. These might arise from the use of non-experimental

data and self-selection or programme placement attributes. It is also necessary todistinguish between before-and-after study and with-or-without enquiry. This study uses the

latter.

Propensity score matching has been for the last decade commonly used to estimate

the impacts of policy and programme interventions (see, for example, Rosenbaum and

Rubin, 1983; Motohashi, 2002; Aert and Czarnitzki, 2004; Criscuolo et al., 2007; Caliendo

and Kopeining, 2008; Mole et al., 2008). Consider the outcome equation below:

y i

= β o

+

Σπ

k

X ik

+ γ BAM i

+ ε i

(1)

i = 1, 2, 3, ...., n; k = 1, 2, 3, ..., m

where y i is a set of outcome variables of firms i ; X

t is a set of observed firm characteristics

influencing their medium and long term growth performance; BAM i represents dummy

membership of firms in BAs; ε i is the randomly distributed error term indicating

the unobservable factors affecting the outcome variables, with zero conditional mean

E (ε i | X

i , M

i ) = 0; π

k and γ are parameters to be estimated.

PSM estimates average treatment effects (ATE) of policy or programme impactsbetween matched treatment and control groups in the region of common support. This is

done by using matching methods such as nearest neighbour (NN) and others69. Treatment

and control groups are matched based on the probability of participation or propensity

score of participation estimated from their observed covariates (Khandker et al. 2010;

Gertler et al. 2011b). Thus, if the independence assumption holds or E (ε i | X

i , BAM

i ) = 0 and

if there is a significant overlap of participants and non-participants in the region of common

support, the average treatment effects can be written as70:

69 Other matching methods include caliper, or radius matching, stratification or interval matching, and kernel

and local linear matching (Khandker et al., 2010).

70 The formula was adapted from Khandker et al., 2010: 57.

k



153

ATE PSM

= [Σ yM –Σ ϕ (i, j ) y N ] (2)

where N m is the number of firms i participating in BAs and ϕ (i, j ) is the weight used to

aggregate outcomes for the matched non-participating firms. However, if zero conditional

mean error assumption is violated or E (ε i | X i , BAM i ) ≠ 0 and the overlapping region of common support is significantly small, PSM would produce inconsistent and biased

coefficients (Caliendo and Kopeining, 2008; Khandker et al., 2010; Gertler et al., 2011a).

We use kernel and nearest neighbour matching methods to estimate the impacts of BA

membership on firms’ performance (Oh et al., 2009).

To obtain reliable, consistent, and unbiased coefficients using PSM, the following

assumptions must be met: (1) the un-confoundedness assumption or the conditional

independence assumption (CIA), where potential outcomes are independent of treatment

assignment given observed covariate x i ; and (2) significant overlapping assumption of common support between participants and non-participants. While the violation of the latter

is less severe; that of the former, so-called selection or programme placement bias, would

be problematic and is technically and practically difficult to address.

Estimation results are more likely to have upward or downward biases due to

administrative and self-selection reasons. If a business association administratively selects

large and already well-established firms as members, assessment results tend to overstate

the benefits received. When small and low productive firms are selected members the

benefits are understated. The study assumes that there is no administrative selection, asalmost all business association membership in almost all business associations in

Cambodia is voluntary.

There is the option of firms choosing to participate in BAs. What encourages

a firm’s participation in BAs is unknown. There could be unobserved characteristics,

affecting outcome variables, which are likely to encourage participation. Some of these

attributes (which may not have available data) are productivity growth, firm owner-manager

ability and advance firms, in terms of managerial, administrative and financial structure.

Results from this are likely to overstate the benefits of business association participation.

The opposite is when participating firms are those with low productivity or low owner-

managers’ ability. This would understate the benefits of BA membership. Our descriptive

statistics (Annex tables 3, 4 and 5) provides some relevant information on this issue. Large

firms, exporting firms, firms with term loans from financial institutions, 100% foreign-owned

and joint venture firms, manufacturing establishments, and firms who know about Special

Economic Zone (SEZ) have higher per centages of being BA members than their

counterparts. Almost all of these firms have higher performance indicators, except labour

productivity and labour cost per worker (Annex table 5).

To tackle selectivity or programme placement or biases of ex-post characteristics,other approaches can be used. This could be difference-in-differences (DID) or

instrumental variables. DID would produce consistent and unbiased coefficients of impacts

only if we control for observable characteristics of firms and the randomly distributed error

1

N M

i ∈M i ∈M i j



154

terms that are not dependent on unobservable attributes. The validity of DID holds if, and

only if unobservable attributes affecting outcome variables are assumed to be constant

over time, However, in practice, this might change. If the error term in DID equation is

time-variant, estimated coefficients would still be biased and inconsistent. To address

selection bias resulting from time-variant observable attributes, the “Instrumental Variables”technique or the two-stage procedure described by Heckman (1979) can be employed.

However, finding good instrumental variable(s) is challenging.

Therefore, PSM is used given the quality of the existing data set. The use of

matching is mainly to balance characteristics of participating and non-participating firms,

rather than to completely eliminate selection bias. Additionally, in constructing a comparison

group for matching, some factors have to be taken into account. First, data on outcome

variables of comparison group, matched with those of treatment group, should be from the

same set of questions (Caliendo and Kopeining, 2008). This condition is met by using the

same set of questions to collect information on participating and non-participating.

Second consideration is the issue of sample size – the total sample size and the

ratio of treatment-to-comparison observations. Basically, matching can be done in two ways

– with and without replacement. The latter approach would require the comparison group to

have a large sample size since one observation of the comparison group can only be

matched one time. Whereas, the former approach would demand a relatively smaller

sample size. The total sample size in our data is 502, and the ratio of participating-to-non-

participating firms is 1:2.6. Since we use matching with replacement (which is common in

matching literature) sample size is not a major issue. Lastly, Lee (2006) pinpoints that CIAwould imply that covariates x

i , needs to be chosen in a way that it is correlated with the

decision to participate and firms’ outcome variables. This is when programme placement is

voluntary. So, observable characteristics that meet the third requirement need to be found.

This can be done by using economic theory or empirical studies.

2. Variables measurement

The choice of variables to be included in subsequent equations is based on

economic theory and previous empirical works. Where there is no economic theory tosupport variable choice, formal statistical tests are used to justify its validity (see, for

example, Heckman et al., 1998; Black and Smith, 2004; Caliendo and Kopeining, 2008).

One of the survey questions is “Is your establishment/firm a member of a business

association or chamber of commerce?” The study uses this as the main binary explanatory

variable, with “1” representing membership and “0” otherwise. Logarithmic form is applied

to the final continuous outcome variables and some indicator variables to make

distributions approximately normal (For example see Roper and Hewitt-Dundas, 2001;

World Bank, 2010; Mole et al., 2008; Motohashi, 2002; Harvie et al., 2010). The definition

of firms is based on the government’s definition. The firm categories are based on the

number of regular employees of the firm in 2006 fiscal year (Annex table 1).



155

The firm’s age is taken from the start of operations in the country to April 2006.

Previous studies argue that firm age is positively related to firms’ high participation

probability in some kind of associative cooperation or production network. This is due to

older firms would have more experience in production, therefore understanding the benefits

they can take from this kind of collective bargaining. Whereas, younger firms may still bedoubtful about the benefits. The study hypothesises that firm age is statistically positively

correlated with high probability of participation. We also introduce square of firm age to

allow for diminishing returns from participation over time.

Firm size is found to be positively related to the decision to participate in BAs.

However, there are arguments about whether small firms are more likely to rely on

intermediary support than larger ones (Salisbury, 1984; World Bank, 2010). Salisbury

(1984: 74) argues that a sector with small firms is more likely to be attracted to associative

bodies where individual small firms can take greater advantage of collective services. This

is because small firms, for example, cannot bear the full costs of private consultancy or

professional services, by themselves. He also states that large firms, that can take

advantage of economies of scale, are more likely to be able to afford private

representations and specialised services. Mitchell (1990: 627-628) shows that large firms

tend to have direct contact with government rather than rely on business or professional

associations for representation. Bennett (1999: 256) also argues that large firms are more

likely to influence government decisions directly rather than rely on the representation of

associative bodies. Bennett (1995: 261) shows that the smallest firms are more likely to

become CoC members than larger ones. Bennett (2000) partly proves Salisbury’s (1984)

argument that the demand for collective services is greater for small firms. Alternatively,

while analysing the impacts of government-supported training programmes in Chile,

Mexico, Colombia and Peru, the World Bank (2010) shows that, common across the

countries, larger firms are more likely to be attracted to support programmes compared to

smaller ones. These different arguments lead us to empirically test the relationship between

firm size and participation probability. This study hypothesises that small firms are more

likely to be attracted to BAs or CoC.

The World Bank (2010) also indicates that more manufacturing SMEs tend to

participate in programmes compared to services and trade sector SMEs. Bennett (1995:269) shows that manufacturing firms are between four or eight times likely to be CoC

members than non-manufacturing firms. In this study, the surveyed firms are categorised

into four sectors – manufacturing, trade, tourism and other (construction, transport, IT and

other). The hypothesis is that manufacturing firms, in general, and SMEs, in particular,

would be more likely to participate in BAs or CoC. The study also hypothesizes that fully

foreign-owned and joint-venture firms and SMEs have a high probability of BA membership

than fully domestically-owned firms. This is because foreign-owned firms may have more

understanding of the benefits they can obtain from the collective services of BAs. Previous

studies have found significant positive relations between exporting firms and SMEs and thedecision to participate in production networks or other kinds of associative bodies. This

study hypothesize that a firm’s decision to become a BA member is higher among exporting

firms and SMEs compared to their non-exporting counterparts.



156

Earlier studies also argue that firms, in general, and SMEs, in particular (with better

access to finance), are more likely to participate in BAs or production networks (Harvie et

al., 2010). Possible credit rationing by financial institutions, resulting from information

imperfections in credit market, can prevent SMEs from accessing term loans or credit line

(Stiglitz and Weiss, 1981). Harvie et al. (2010) indicates that easier access to finance or high financial leverage increases the probability of SMEs’ participation in production

networks. However, this study hypothesize that SMEs are more likely to participate in BAs

so that they can use membership as some sort of guarantee when they approach banks for

term loans or credit line. This is so for SMEs that finds it difficult to access finance due to

size constraint and inability to provide required collateral to banks. The study also includes

other variables that may have an influence on participation in the selection equation.

C. Data source and descriptive statistics

1. Data source

The study uses the 2007 World Bank Enterprise Surveys of Cambodia. A simple

random sample from a combined sampling frame (World Bank, 2007) was employed to

select the required observations from each strata defined by four sectors (manufacturing,

tourism, trade, and others), as well as three firm sizes. With the elimination of surveyed

firms that had less than five regular employees in April 2006, the total sample size was 502,

204 of which have between 5 and 19 workers, 146 have between 20 and 99 workers, and

152 have more than 99 workers. Based on firm categories defined by the government andthe number of regular employees in 2006 fiscal year (Annex table 1), there were 361

MSMEs (72 per cent of the total sampled firms) while 141 large enterprises accounted for

the remaining 28 per cent.

The survey covered firms that started operating before April 2006 in a number of

sub-sectors ranging from textiles, garments, wholesale (including export services) to hotels

and restaurants. Surveyed firms are categorised into four sectors – manufacturing, trade,

tourism and other (including construction, transport and IT). The survey considers only

private-for-profit firms, not government-owned or community-owned establishments. Also,firms who did not keep their own accounts were eliminated. The survey took place in four

main areas – Battambang, Siem Reap, Phnom Penh and Kampong Cham. Most surveyed

firms (401 or 80 per cent of total sample firms) were located in Phnom Penh and its

outskirts in the Kandal province. The remaining firms are in Battambang (9 or 2 per cent),

Siem Reap (66 or 13 per cent), Kampong Cham (6 or 1 per cent) and “others” (20 or 4 per

cent).

2. Descriptive statistics

Of the total sampled firms, 139 (27.7 per cent) were members of a BA or CoC, while

the remaining 363 firms were not. Of the 363 MSMEs, only 47 were members of a BA while

the other 314 were not. There were 92 large firms that reported membership, while the

other 49 did not. Only 33.8 per cent of MSMEs are members of a BA or CoC compared with



157

66.2 per cent of large firms, which are mainly garment enterprises. Most studies found that

10 per cent, or even less, of all SMEs are members of associations. This low membership

may be due to some of the frequently observed characteristics of MSMEs, in particular

family-based proprietorship and the belief of SME owner-managers, that they would have

little to gain from an association’s collective or specific activities. Large firms, foreign and joint venture firms, manufacturing establishments, firms with large national market share,

firms where majority of workers are union members, firms that have more foreign

competitors, firms that have term loans with financial institutions, firms in special economic

zones (SEZ) tended to have higher levels of membership in BA or CoC (Annex table 4).

Annex Table 5 provides statistical relationships between a number of firms’

characteristics and outcome indicators. ANOVA is used to test mean differences of growth

performance and each category of characteristics. SMEs and large firms perform

comparatively better than micro firms, having more sales, higher production and labour

productivity and paying higher total and per worker costs. Large firms have higher levels of

outcome variables than SMEs but lower labour productivity and labour cost per worker. The

differences are statistically significant at 5 per cent. This is interesting in the sense that

large firms, mainly firms in the garments sector, produce more simply because they have

more workers. However, each worker is less productive compared to an employee working

in SMEs. Other relationships between firms’ characteristics and performance also show

similar results. With regards to what services participating firms expect from BAs,

information and/or contacts on domestic product and input markets is important for micro,

small and medium enterprises but less important for large firms (Annex table 6).

Annex Table 8 reports mean differences of a number of firms’ outcome indicators

(in natural logarithms) and other characteristics between members and non-members of

BAs. Mean values of outcome indicators are higher and almost all of these differences are

statistically significant. Outcome indicators are annual sales, production, costs of labour,

cost of intermediate goods, other costs, labour productivity, and cost per worker of firms

participating in BAs or CoC. These results tend to provide a preliminary indication that BA

membership has positive impacts on participating firms compared to their non-participating

counterparts.

Also, firms with a high-educated workforce tend to be able to self-operate and

improve their performance without assistance from BAs. The World Bank (2010) found

mixed results on mean differences of programme participation in Chile in 2004.71 It points

out that this could indicate that programmes are more attractive to poorly performing firms

and that participation improves their performance relative to what it might have been if they

had not participated.

The statistically positive significance (1 per cent level) of mean differences of major

outcome indicators still holds for the sub-sample of SMEs (Annex table 9). Participating

SMEs appear to benefit from the services of BAs or CoC, given their high mean values of

71 But note that there was no controlling for other contributing factors.



158

outcome indicators compared to non-participating SMEs. Approximately 74.85 per cent

of surveyed firms are non-exporters while only 25.15 per cent are exporting firms. The

Chi2-square test of independence between exporters and non-exporters and firms’ decision

to become BA member indicates that 63.5 per cent of exporting firms are members of BAs

or CoC, while 36.5 per cent are not. Of the non-exporters, 15.5 per cent are reportedly BAmembers while 84.5 per cent are not. The difference is statistically significant at 1 per cent

level, signifying that exporting firms have a higher probability of participation in BAs than

non-exporting ones.

More manufacturing firms (62.7 per cent) are members of BAs compared to firms in

the trading sector (7.6 per cent), tourism sector (16.8 per cent) and other sectors (19.3 per

cent). The Chi2 square test of independence between row and column is again statistically

significant at 1 per cent level. A large percentage of firms in Phnom Penh (31.2 per cent)

are members of BAs compared to Siem Reap (16.7 per cent), Kampong Cham (16.7 per

cent) and others (2 per cent). This partly reflects associative organizations being more

concentrated in Phnom Penh. As far as equity stakes of firms are concerned, a high

percentage of firms that are fully foreign-owned (52.1 per cent) and joint venture (52.1 per

cent) are BA members. Only 13.8 per cent of fully domestically-owned firms are members.

D. Regression results and discussion

1. Participation likelihood

Annex table 10 presents results of logistic regression analysis, predicting the

likelihood of firms’ participation in BA or CoC given their observed covariates. Regressions

are run for both the pooled sample (all participating and non-participating firms) and a sub-

sample (participating and non-participating MSMEs). In estimation, six different model

specifications are used, partly to test the robustness of results. Results reveal that most of

the covariates have expected effects (in terms of the sign) on participation probability,

though with varying statistical significance.

For the pooled sample, the variable “years in operation” is positively correlated

statistically with high probability of participation. This indicates that older firms tend to

participate in BA or CoC more than younger ones. Second, large firms are more likely to

register membership than MSMEs. This finding contradicts our prior expectation that

small firms would have higher participation likelihood as they can use membership as

a bargaining tool for business survival and growth. However, the result is consistent with

that of previous studies (World Bank, 2010; Harvie et al., 2010).

Non-exporting firms accounted for 74.8 per cent of the total sample, compared to

25.2 per cent of exporting firms. Yet, exporters, who also tend to be fully foreign-owned or

joint ventures, are more likely to participate than non-exporting ones. This may be because

exporting firms can get business information on foreign markets and advice and support

services from BA or CoC (Spence, 2003). It should be emphasised that more needs to be

done to encourage domestic producers to be exporters. Of fully domestically-owned firms



159

78.4 per cent are non-exporters, and 92.3 per cent of output is sold domestically compared

to 6.2 per cent that are exported directly and 1.5 per cent that are exported indirectly (World

Bank 2007).

The large proportion of non-exporting domestic firms may indicate several possible

constraints to exporting (1) long and complicated administrative compliance, (2) limited

access to information on foreign markets, (3) products are low quality and do not meet

internationally required standards, and (4) limited staff capacity to go international. The

problems reported by exporting firms in the survey five years ago remain today. These are

lengthy and costly export procedures, informal fees, gifts giving to officials and high cost

per export container. Cambodia was ranked 138 out of 183 economies evaluated on ease

of conducting business (World Bank, 2012).

Also, more needs to be done to reduce export regulatory and administrative

compliance costs. Almost all firms in Cambodia are MSMEs, where production capacity isstill low and vulnerable to competition from imports. High costs of doing business also force

MSMEs to stay small and informal, further jeopardising their expansion and productivity.

Business management structure is predominantly conventional, with individual and/or

family as the largest shareholder or owner. Data shows that 49.8 per cent of firms reported

having an individual as the largest shareholder or owner, and 32.4 per cent having family

as the largest shareholder. One of the risks of family-based firms dominating the business

environment is resistance to change in management style, and reluctance to adopt new

technologies.

There is merit in pushing domestic producers to eye foreign markets. Previous

studies have also documented the need for domestic firms to upgrade their production

chain and the potential benefits of exploring opportunities outside home markets. Using

Colombian panel data on manufacturing firms, Isgut (2001) finds that exporters tend to be

larger, more productive, pay higher wages, are more capital intensive, and have higher

labour productivity than non-exporters. Biesebroeck (2003) shows that exporting

manufacturing firms in Sub-Saharan African countries tend to be more productive than

non-exporters. This is due to exporting firms improving their productivity upon entering

foreign markets. He argues that scale economies can be one of the determining factors.Bernard and Jensen (1997) also found that exporting manufacturing firms contributed

significantly to the demand for skilled labour and wage increases in US manufacturing

plants during the 1980s.

Thus, encouraging domestic producers to export means helping them to get out of

their comfort zone and start exploring productivity and profitability opportunities beyond

domestic markets. This is especially relevant to Cambodian producers given the intensity of

regional and global economic integration that Cambodia is currently taking part in. The

effects of the Asian Free Trade Agreement, specifically ASEAN Economic Community,

demand further changes in current production practices, management and innovation

among domestic producers. This is to become competitive and to stay afloat when

domestic markets are opened to an influx of foreign goods and, at a later stage, freer flow

of services. The government’s current commitment to achieve exports of one million tonnes



160

of milled rice by 2015 is a starting point to improve production capacity and productivity of

domestic rice producers (CEFP, 2011). But, entry into exporting by itself is not enough to

enhance productivity, since competitiveness depends on a number of interdependent

factors in the whole value chain. A long-term strategy is needed to improve infrastructure,

logistic systems, information flow and market access, banking systems, governmentregulations and procedures and human capital. Additionally, support and initiatives should

not be confined to the rice sector as sectoral diversification is a major contributing factor to

fast and sustained economic growth.

Manufacturing firms have higher participation probability than firms in trade, tourism

and other sectors. This provides an early indication that further efforts should be extended

to cultivate the idea of forming more concentrated and specialised form of associative

bodies. This could be in the form of industrial clusters among manufacturers in order to

gain greater benefits, rather than business information only. The higher its share of the

national market, the more likely it is that a firm will participate in BA or CoC. This may be an

indication that firms would use membership to better understand market potential and

further expand their market share.

We hypothesize that firms that have difficulties in obtaining term loans from financial

institutions are more likely to participate in BA or CoC. Nonetheless, the study’s logistic

regression estimates of the six specifications reveals that firms with term loans have

a higher likelihood of participating than those who do not, and it is statistically significant at

1 per cent level. This could be due to a number of reasons. First, becoming a member of an

associative organization is not enough in itself to secure term loans from financialinstitutions. In other words, financial institutions need hard and valuable collateral to guard

against risk of default. Second, this can reflect the fact that BA members are

well-established firms that have a large number of employees, a big share of domestic

markets, and high productivity.72

This finding, however, is consistent with that of previous studies (Harvie et al., 2010;

Harner, 2003; Dinh et al., 2010; Ayyagari et al., 2008). Recent literature points to the lack of

access to financial services or insufficient functioning of the banking system as one of the

major constraints to growth of both large and small firms. In the Cambodian context, whereregulatory and law enforcement is weak and overall public trust low, achieving sufficient

access to formal financing has been and will continue to be a challenge.

The data shows that only 21.7 per cent of surveyed firms reported having term

loans (more than 6 months) with financial institutions. This reflects the relatively low

utilization of formal financing. From the demand side, firms request financial institutions to

increase loan amounts, extend maturity date of term loans and importantly reduce interest

rates. From the supply side, banks and other financial institutions usually raise the issue of

protecting depositors by screening and targeting the best borrowers, applying strict

72 This raises the possibility of endogeneity bias in the estimation, though the PSM method can minimize the

problem, given its balancing of treatment and control groups.



161

requirements to fulfil bank procedures. They also raise the issue of the informal nature of

the business undertakings of most MSMEs which do not have appropriate financial

statements or proper accounting systems. Harner (2003) found that although the demand

for medium and long-term loans is high among SMEs in Cambodia, banks are reluctant to

lend because (1) current contract and law enforcement is weak, (2) high interest rates ondeposits, (3) banks’ limited access to long-term capital, (4) unavailability and difficulty

gathering information on borrowers to conduct credit worthiness analysis, (5) high liquidity

ratios required by the National Bank of Cambodia, and (6) limited staff capacity.

Firms where majority of employees are members of a workers union tend to

participate more than firms with fewer employees belonging to the union. This may be

because associations can act collectively to efficiently bargain and meet union demands.

Recently, there have been protests and strikes by union-member workers, especially in the

garment and footwear industries. They were demanding better wage increases, respect for

labour law and good working conditions. But finding solutions has been difficult. Relations

between employers, the union and the Garment Manufacturers Association of Cambodia,

have soured. Although the right to form associative bodies, like a workers’ union, is an

integral part of Cambodia’s law, there are complaints about the large number of unions in

Cambodia. The study also finds that the number of licenses, permits and regulations that

firms are supposed to meet is highly positively correlated with probability of participation.

The study finds that firms who reported knowing about the existence and functioning of

special economic zones (SEZs) are more likely to be members. This may imply that BAs or

CoC should disseminate more information about the benefits members can get from

participating.

For the sub-sample of MSMEs, most estimation results were similar to those from

the pooled sample.

2. Average treatment effects (ATE)

Once the propensity score is estimated using logistic regression, two commonly

used matching methods are employed – nearest neighbour and kernel. This is to estimate

ATE of programme participation. Matching methods use the estimated propensity score of each specification. The nearest neighbour (NN) matching estimates the effects of

membership participation. This compares participating firms in the treatment group with

non-participating firms in the control group that have similar propensity score. NN matching

is often used with small sample size (World Bank, 2010). One of the weaknesses of NN

matching is that matching quality can be affected if many firms in the treatment group have

high propensity scores, while only a few firms in the control group have high propensity

scores. This shortcoming is addressed by using the kernel matching method. It should also

be noted that the comparison was performed on participating and non-participating firms

who fell in the region of common support. All of the study specifications satisfied thebalancing property (Annex table 10). ATE is estimated for the pooled sample of all firms

and the sub-sample of MSMEs.



162

Annex table 11 reports the ATE estimates for the pooled sample. It shows that

participating firms generally have higher mean value of final outcome. However, these are

statistically significant only for some outcome variables and the signs are not always as

expected. Relative to their matched non-participating firms, firms in business associations

tended to have higher turnover and production, and to spend more on production and other related costs. But, membership has no significant impact on firms’ labour productivity and

labour cost per worker. This may be due to two reasons (1) BA still has limited capacity to

enhance productivity of members through quality training, and (2) large firm participation is

not motivated by the desire to use the association’s training services for the improvement of

productivity, but for various other benefits.

For the sub-sample of MSMEs, ATE estimates are broadly similar to results from

the pooled sample (Annex table 12). Again, membership does not guarantee higher

productivity and cost per worker. Matching results also suggest that MSMEs may benefit

more from participation than large firms.

It must be emphasised that the positive association between membership

participation and firm performance outcomes does not imply causality between participation

and high performance. We do not have baseline survey information on the level of outcome

variables prior to participating. However, the positive trends are largely consistent with

previous studies. Positive trends are also consistent with the pluralists’ view that a business

association (as a collective body promoting efficiency of members’ businesses), is not

merely a rent-seeking and special interest group, as claimed by public choice theorists.

E. Limitations

Analysis and, to some extent, results of the study are subject to some important

limitations imposed by data constraints. First, the cross-sectional data employed could not

provide a full picture of changing characteristics and dynamism of firms over time. That

would require panel data. Second, since baseline characteristics of firms are not available,

estimating the impacts of BAs on firms’ final outcome variables using PSM needs to be

regarded with caution (Gertler et al., 2011b: 115).

Conclusion and areas for future research

This study investigates how a business association or a chamber of commerce in

Cambodia can assist its members to improve productivity and enhance firms’ performance.

A growing number of such bodies have been established with a view to promoting

members’ collective voice and bargaining power. With the emergence of larger enterprises

and a number of MSMEs, these associations may be more important in Cambodia than

ever. However, BAs are still fragmented and membership awareness is still relatively low.Firms are more inclined to operate independently because they are still sceptical about

collective activities. Also, associations face challenges in terms of resources, size, cost

effectiveness, human capital and scope and coverage of services.



163

The study finds evidence that participation in business association tends to be

positively associated with better performance across several outcome variables. This is

done using propensity score matching, ordinary least squares and a combination of PSM

and OLS methods with firm-level cross-sectional data from 2007. These results are

stronger for micro, small and medium enterprises. This suggests that business associationsmay be more beneficial for smaller firms. However, membership appears to have no

significant positive impact on labour productivity and payment per worker.

The study suggests that firms, particularly micro enterprises and SMEs, do benefit

from collective bodies. These participating firms can take advantage of market information,

knowledge sharing and updates on government administrative and regulatory

requirements. But there needs to be more study to assess the benefits of BAs services.

Also needing more attention, is how these services contribute to further growth and

competitiveness of SMEs. Lastly, consideration must be given to whether membership

should be voluntary or obligatory, or a mixture of both.



164

Table 2. Distribution of participating and non-participating groups

by firm size and sub-sector

Firm Size and BA participation

Sub-sector Micro Small and medium Large

Participant Non- Participant Non- Participant Non-

participant participant participant

Food 0 0 4 5 1 1

(0.0) (0.0) (9.5) (2.7) (1.1) (2.0)

Textiles 0 0 0 1 5 0

(0.0) (0.0) (0.0) (0.5) (5.4) (0.0)

Garments 0 0 1 2 68 21

(0.0) (0.0) (2.3) (1.1) (74.0) (42.8)

Plastics & Rubber 0 1 0 2 2 1

(0.0) (0.7) (0.0) (1.1) (2.2) (2.0)

Basic metals and ... 1 ... 2 ... 0

fabricated metal products (0.7) (1.1) (0.0)

Other manufacturing 1 3 0 8 2 2

(20.0) (2.3) (0.0) (4.5) (2.2) (4.1)Wholesale (including 1 3 6 21 1 2

export services) (20.0) (2.3) (14.3) (11.7) (1.1) (4.1)

Retails ... 44 ... 25 ... 2

(33.1) (13.8) (4.1)

Hotels & Restaurants 1 30 12 62 5 9

(20.0) (22.6) (28.6) (34.2) (5.4) (18.4)

Other services 2 16 4 3 0 0

(travel agencies, tour (40.0) (12.0) (9.5) (1.6) (0.0) (0.0)

operators, etc.)Construction 0 2 2 4 1 0

(0.0) (1.5) (4.8) (2.2) (1.1) (0.0)

Annex

Table 1. Classification of enterprises in Cambodia

Types Employees (persons) Start-up capital (USD)

Micro-enterprises Fewer than 10 Less than 50,000

Small Between 11 and 50 Between 50,000 and 250,000

Medium Between 51 and 100 Between 250,000 and 500,000

Large 100 or more 500,000 or more

Source: RGC (2005: 13)



165

Transport 0 7 5 13 1 0

(0.0) (5.3) (11.9) (7.3) (1.1) (0.0)

IT ... 3 ... 3 ... 0

(2.3) (1.6) (0.0)

Other 0 23 8 30 6 11

(0.0) (17.2) (19.1) (16.6) (6.4) (22.5)

Total 5 133 42 181 92 49

(100.0) (100.0) (100.0) (100.0) (100.0) (100.0)

Source: Author’s calculation

Figures in brackets are column percentages of the number of firms reporting membership of BA or CoC.

Table 3. Number of establishments as exporter

Establishment Micro Small and medium Large Total

sector

Exporter 11 18 97 126

(8.7) (14.3) (77.0) (100.0)

Non-exporter 127 204 44 375

(33.9) (54.4) (11.7) (100.00)

Total 138 222 141 501

(40.52) (29.14) (30.34) (100.00)


Figures in brackets are row percentages of the total.


Firm Size and BA participation

Sub-sector Micro Small and medium Large

Participant Non- Participant Non- Participant Non-

participant participant participant



166

Table 4. Membership in business association or chamber of commerce

by firm’s characteristics

Firm’s characteristicsMembership in BA or CoC

Yes No Total obs.Firm size

Micro (employees <10) 5 133 138

(3.6) (36.5) (27.5)

Small and medium 42 181 223

(11< = employees < = 100) (30.2) (49.9) (44.4)

Large (employees >100) 92 49 141

(66.2) (13.6) (28.1)

Exporter

Exporting firm 80 46 126

(58.0) (12.7) (25.2)

Non-exporting firm 58 317 375

(42.0) (87.3) (74.8)

Firm with term loans

Firm having term loans with financial 40 69 109

institutions (28.8) (19.0) (21.7)

Firm having no term loans with financial 99 294 393

institutions (71.2) (81.0) (78.3)

Firm’s equity stake

100% domestically owned 44 275 319

(31.7) (75.8) (63.6)

100% foreign owned 74 68 142

(53.2) (18.7) (28.3)

Joint venture 21 19 40

(15.1) (5.5) (8.1)

Firm typeManufacturing 84 50 134

(60.4) (13.7) (26.7)

Trade 8 97 105

(5.8) (26.7) (20.9)

Tourism 24 120 144

(17.3) (33.1) (28.7)

Other 23 96 119

(16.5) (26.5) (23.7)Geographical location

Battambang 0 9 9

(0.0) (2.5) (1.8)



167

Siem Reap 11 55 66(7.9) (15.2) (13.2)

Phnom Penh 125 276 401

(89.9) (76.0) (79.9)

Kampong Cham 1 5 6

(0.7) (1.3) (1.1)

Other 2 18 20

(1.5) (5.0) (4.0)

Firm age0-4 50 170 220

(36.0) (46.8) (43.7)

5-10 67 120 187

(48.2) (33.1) (37.3)

More than 10 22 73 95

(15.8) (20.1) (19.0)

# of competitors as domestic firms

None 35 16 51(25.8) (4.5) (10.4)

1 2 4 6

(1.5) (1.1) (1.2)

2-5 21 37 58

(15.7) (10.3) (11.7)

>5 77 302 379

(57.0) (84.1) (76.7)

Firm’s information about SEZYes 97 137 234

(70.0) (37.7) (46.6)

No 42 226 268

(30.0) (62.3) (53.4)


a) Figures in brackets are column percentages of the number of firms reportingmembership of BA or CoC.

b) Pearson Chi2 was conducted to test column differences of each firm’s characteristic on

the percentage of firms reporting membership of BA or CoC. All tests are significant, atleast, at 5% level.


Firm’s characteristicsMembership in BA or CoC

Yes No Total obs.



168

Table5.Firm’scharacte

risticsandperformance

Turnover

Produ

ction

Labour

Cos

tof

Labour

Other

#of

(log)

(log)

productivity

labour

costper

cost

employed

Firm’scharacteristics

(log)

(lo

g)

worker

(log)

workers

(log)

(log)

Firmsize

Micro(employees<10)

10.816

10.36

7

8.463

8.716

6.806

22.401

1.906

(1.267)***

(1.30

0)***

(1.301)***

(.856)***

(.851)***

(6.952)***

(.229)***

Smalland

medium(11<=employees<=

100)

12.721

12.34

6

8.984

10.607

7.258

29.491

3.347

(1.530)***

(1.45

9)***

(1.337)***

(1.078)***

(.826)*

(8.159)***

(.612)***

Large(employees>100)

15.759

15.02

0

8.637

13.519

7.147

40.837

6.369

(1.434)***

(1.41

6)***

(1.257)*

(1.037)***

(.850)*

(7.383)***

(.885)***

Exporter

Exportingfirm

15.403

14.67

0

8.766

12.937

7.046

38.849

5.859

(.176)***

(.17

1)***

(.129)

(.160)***

(.077)

(.769)***

(.161)***

Non-exportingfirm

12.301

11.88

6

8.738

10.241

7.122

27.991

3.110

(.101)

(.10

1)

(.069)

(.088)

(.044)

(.484)

(.062)

Firmwithtermloans

Firmwithtermloans

13.449

12.97

9

9.039

11.135

7.211

31.152

3.905

(.220)**

(.21

3)**

(.135)**

(.190)

(.086)

(1.053)

(.156)

Firmwitho

uttermloans

12.969

12.48

7

8.667

10.860

7.075

30.612

3.770

(.121)

(.117)

(.068)

(.106)

(.043)

(.511)

(.094)



169

Table5. (

c o n t i n u e d )

Turnover

Produ

ction

Labour

Cos

tof

Labour

Other

#of

(log)

(log)

productivity

labour

costper

cost

employed


(log)

(lo

g)

worker

(log)

workers

(log)

(log)

Firm’sequity

stake

100%dom

esticallyowned

12.019

11.64

3

8.630

9.976

6.989

26.259

2.975

(1.796)***

(1.77

4)***

(1.246)

(1.567)***

(.798)***

(8.313)***

(1.211)***

100%foreign-owned

14.955

14.28

2

8.853

12.653

7.238

39.297

5.382

(2.110)***

(1.99

2)***

(1.433)

(1.859)***

(.954)***

(8.059)***

(1.878)***

Jointventu

re

14.783

14.18

5

9.342

12.407

7.568

36.046

4.759

(1.950)***

(1.96

8)***

(1.353)***

(1.499)***

(.775)***

(10.462)***

(1.577)***

Firmtype

Manufactu

ring

15.183

14.47

6

8.596

12.86

16.961

37.725

5.899

(2.073)***

(1.92

8)***

(1.201)

(1.772)***

(.765)

(10.335)***

(1.638)***

Trade

12.176

11.71

5

9.107

9.704

7.101**

26.396

2.604

(2.044)

(1.99

8)

(1.535)***

(1.309)

(.678)

(8.228)

(.860)

Tourism

12.116

11.72

0

8.518

10.081

6.923

27.926

3.151

(1.659)

(1.68

2)

(.952)

(1.686)

(.863)**

(8.628)

(1.138)

Other

12.630

12.19

7

8.885

10.794

7.495

30.065

3.275

(2.155)

(2.05

3)

(1.547)

(1.804)

(.972)

(9.974)

(1.358)



170

Table5. (

c o n t i n u e d )

Turnover

Produ

ction

Labour

Cos

tof

Labour

Other

#of

(log)

(log)

productivity

labour

costper

cost

employed


(log)

(lo

g)

worker

(log)

workers

(log)

(log)

Geographica

llocation

Battamban

g

11.446

11.22

1

8.269

9.230

6.278

21.682

2.951

(1.884)

(1.81

6)

(.963)

(2.261)

(1.218)

(8.290)

(1.465)

SiemReap

12.393

11.97

4

8.530

10.454

7.032

28.272

3.422

(1.636)

(1.71

6)

(1.009)

(1.610)

(.718)

(7.068)

(1.214)

PhnomPe

nh

13.310

12.80

7

8.822

11.111

7.159

31.720

3.930

(2.423)

(2.27

6)

(1.383)

(2.101)

(.873)

(10.528)

(1.909)

KampongCham

12.066

12.19

8

9.319

9.299

6.574

25.171

2.723

(2.778)

(2.71

6)

(1.221)

(2.142)

(.571)

(11.172)

(1.776)

Other

11.723

11.22

1

8.083

9.920

6.782

24.702

3.138

(2.129)

(2.07

2)

(.797)

(1.876)

(.561)

(10.627)

(1.548)

Firm’sinform

ationaboutSEZ

Yes

13.897

13.32

9

9.123

11.451

7.273

33.623

4.171

(.146)***

(.14

3)***

(.096)***

(.130)***

(.056)***

(.653)***

(.123)***

No

12.356

11.95

6

8.422

10.453

6.957

28.202

3.476

(.139)

(.13

6)

(.071)

(.125)

(.051)

(.607)

(.104)

S o u r c e :

Auth

or’scalculation.Figuresinbracketsarestandarddeviationsexceptfiguresforexporterandfirm’sinformationonSEZ,whichare

standarderrors.

***indicates1%significantlevel,**5

%level,*10%level.



171

Table 6. Perceived important services members expect from BA or CoC

(1) (2) (3) (4) (5) (6)

Micro (employees < 10) 3 2 5 4 4 4

(3.1) (2.1) (6.2) (5.1) (4.8) (3.5)Small and medium 25 19 25 22 24 34

(11 < = employees < = 100) (26.0) (20.0) (30.9) (28.2) (28.6) (30.1)

Large (employees >100) 68 74 51 52 56 75

(71.0) (77.9) (62.9) (66.7) (66.7) (66.4)

Pearson Chi2 of column 2.9939 18.4378*** 3.9121 1.4023 0.9981 0.0119

differences


a) Services enterprise received from business association that are most important, includes(1) lobbying government, (2) resolution of disputes (with officials, workers or other firms),(3) information and/or contacts on domestic product and input markets, (4) informationand/or contacts on international product and input markets, (5) accrediting standards or quality of products and reputational benefits, and (6) information on governmentregulations.

b) Figures in brackets are column percentages of the number of firms reporting theimportance of each service.

c) *** indicates 1% significant level, ** 5% level, * 10% level.

Table 7. Firm’s years of experience since establishment and performance

Firm’s years Turnover Production Labour Cost of Labour Other # of

of experience (log) (log) productivity labour cost per cost employed

(log) (log) worker (log) workers

(log) (log)

1 12.990 12.549 8.487 10.819 6.871 31.709 3.915

(2.503) (2.243) (1.316) (2.069) (.839) (11.189) (1.948)

2 12.995 12.557 8.651 10.960 7.090 29.912 3.852

(2.172) (2.103) (1.295) (2.035) (.691) (10.199) (1.751)

3 12.864 12.413 8.560 10.866 7.061 31.681 3.805

(2.210) (2.130) (1.370) (2.005) (.684) (8.583) (1.814)

4 12.384 12.023 8.934 10.260 7.223 28.760 3.036

(1.979) (1.843) (1.065) (1.913) (.940) (8.097) (1.442)

5 13.040 12.807 8.792 11.342 7.275 31.994 3.869

(2.156) (2.172) (1.208) (1.971) (.858) (10.157) (1.771)

6 12.907 12.437 8.766 10.490 6.880 31.265 3.610

(2.294) (2.190) (1.534) (1.874) (.855) (10.006) (1.646)

7 13.572 12.732 8.497 11.326 6.965 34.127 4.360(2.479) (2.482) (1.110) (2.189) (.893) (8.790) (2.070)

8 14.126 13.390 8.884 11.753 7.247 32.156 4.506

(2.632) (2.430) (1.127) (2.095) (1.004) (10.685) (2.030)



172

9 13.740 13.114 8.593 11.597 7.064 32.729 4.447

(2.779) (2.663) (1.399) (2.536) (.883) (12.411) (2.170)

>10 13.011 12.554 9.001 10.861 7.312 28.557 3.568

(2.289) (2.247) (1.380) (2.028) (.868) (10.635) (1.659)

Prob >F 0.1091 0.424 0.3013 0.0451** 0.0187** 0.1438 0.0092***

Adj R-Squared 0.0112 0.003 0.0036 0.0168 0.0168 0.0090 0.0257

Obs. 486 466 466 494 494 502 502

Bartlett’s test for 0.6250 0.7130 0.4350 0.8480 0.3410 0.1910 0.2430

equal variance

(Prob > Chi2)


a) Firm’s years of experience are calculated by subtracting the year of establishment andsurvey year, which is 2006.

b) Figures in brackets are standard deviations.

c) *** indicates 1% significant level, ** 5% level, * 10% level.


Firm’s years Turnover Production Labour Cost of Labour Other # of

of experience (log) (log) productivity labour cost per cost employed

(log) (log) worker (log) workers

(log) (log)

Table 8. Descriptive statistics of main outcome and indicator variables for all firmsa

Non-participating Participating Mean T-

Indicator variables firms firms difference statistics

N Mean N Mean Diff t

Firm age 363 6.57 139 6.47 -0.01 -0.21

Sales (log)

2005 325 12.13 124 15.03 2.90 13.57***

2006 350 12.29 136 15.09 2.81 13.92***

Production (log)b

2005 309 11.72 116 14.42 2.70 12.76***

2006 334 11.84 129 14.46 2.62 13.65***

Cost of labour (log)

2005 326 10.15 125 12.68 2.53 13.99***

2006 355 10.92 137 12.77 2.58 14.99***

Cost of intermediate

goods (log)

2005 331 10.67 126 13.54 2.89 12.35***

2006 358 10.86 138 13.60 2.74 12.31***

Other costs (log)c

2005 329 9.68 123 11.96 2.28 14.07***

2006 354 9.84 134 12.08 2.24 15.89***



173





Productivity (production/labour)

(log)

2005 304 8.59 118 9.11 0.52 3.64***

2006 332 8.66 128 8.91 0.25 1.94***

Labour cost per worker (log)

2005 325 7.11 124 7.22 0.11 1.20

2006 347 7.09 132 7.27 0.17 2.21**

Number of permanent workers

(log)

2005 344 2.97 130 5.39 2.41 14.51***

2006 359 3.09 139 5.52 2.43 15.26***

Total permanent skilled 171 67.48 86 60.34 -7.15 -1.39

employees received

training (%)

Foreign nationals of the total 355 0.75 137 8.71 7.97 7.35***

Number of customers (log) 249 1.77 58 1.89 .12 1.59

Senior management time

(in dealing with government

requirement) (%)

2004 297 8.81 116 12.11 3.29 1.50

2006 359 8.79 139 13.20 4.41 2.18***

Educational levels of

employees

Primary school 346 15.61 139 37.46 21.76 7.27***

(below grade 6) (%)

Up to Lower Secondary 360 25.67 136 21.15 -4.53 -2.13***

(grade 7-9) (%)

Up to Upper Secondary 359 27.42 135 18.77 -8.65 -4.01***

(grade 10-12) (%)Up to Universities & 360 28.053 139 19.316 -8.74 -2.85***

institutions (%)


a) Independent t-tests are used to test the mean differences of a number of outcomevariables (in logarithms), other characteristics of member firms in BAs and those of non-member firms. Bartlett’s test of the null hypothesis of equal variance is calculated.

b) Other costs of production include costs of electricity, fuel, water, communication services,transportation, rents, and repairs and maintenance. Mean differences are the differencesbetween outcome indicators of firms who are members of BAs or Chamber of Commerceand those of firms who are non-members.

c) Values of total sales, production and productivity in 2005 are inflated using Consumer Price Index calculate by the National Institute of Statistics, Cambodia. In 2005, annualaverage CPI is 115.20 and 120.63 in 2006. July-December 2000 is based year.

*** indicates that the mean differences are significant at 1% level, ** at 5% level and* 10% level.



174

Table 9. Descriptive statistics of main outcome and indicator variables for SMEs




Firm age 307 6.73 43 7.37 0.64 0.87

Sales (log)

2005 260 11.61 39 13.23 1.62 5.97***

2006 295 11.75 41 13.20 1.44 5.46***

Production (log)

2005 260 11.20 39 12.79 1.59 5.88***

2006 280 11.38 40 12.76 1.38 5.12***

Cost of labour (log)

2005 277 9.65 40 10.74 1.16 4.84***

2006 299 9.69 43 10.81 1.11 5.26***

Cost of intermediate

goods (log)

2005 279 10.14 41 11.85 1.71 5.55***

2006 299 10.34 43 11.87 1.54 5.03***

Other costs (log)

2005 279 9.28 40 10.68 1.39 6.50***

2006 300 9.46 42 10.80 1.34 6.58***

Productivity

(production/labour) (log)

2005 255 8.54 39 9.40 0.86 3.41***

2006 272 8.57 40 9.33 0.76 3.24***

Labour cost per worker (log)

2005 266 7.06 40 7.35 0.28 1.60

2006 292 7.07 43 7.37 0.31 1.87*

Number of permanent

workers (log)

2005 296 2.57 42 3.23 0.65 4.73***

2006 307 2.67 41 3.36 0.68 5.13***

Total permanent skilled 142 66.88 29 63.48 -3.40 -0.44

employees received

training (%)

Total permanent unskilled 142 41.69 28 31.89 -9.79 -1.03

employees received

training (%)

Foreign nationals of

the total (%)

Number of customers (log) 223 1.75 34 1.98 0.22 2.52***



175

Educational levels

of employees

Primary school 304 16.62 43 14.65 -1.97 -0.50

(below grade 6)

Up to Lower Secondary 304 26.12 43 17.34 -8.77 -2.57***

(grade 7-9)

Up to Upper Secondary 304 27.73 43 32.41 4.68 0.99

(grade 10-12)

Up to Universities & 304 29.51 43 35.58 6.06 1.04

institutionsSource: Author’s calculation

a) Other costs of production include costs of electricity, fuel, water, communication services,transportation, rents, and repairs and maintenance. Mean differences are the differencesbetween outcome indicators of SMEs who are members of BAs or CoC and those of SMEs who are non-members.

b) Values of total sales, production and productivity in 2005 is inflated using Consumer Price Index calculate by the National Institute of Statistics, Cambodia. In 2005, annualaverage CPI is 115.20 and 120.63 in 2006.

*** indicates that the mean differences is significant at 1% level, ** at 5% level and * 10%level.







176

T

able10.Logitregressiontoestimatepropensityscor

eofmembershipparticipationconditionalonobserve

d

andselecte

dcovariates

C

ovariates

(1)

(2)

(3)

(4)

(5)

(6)

Pooled

MSME

Pooled

MSME

Pooled

MSME

Pooled

M

SME

Pooled

MSME

Pooled

MSME

Dependentvariable:participation

inbusinessas

sociationorchambers

ofcommerce(

1=participate)

Independentv

ariables


Firmsinopera

tion(years)

.065*

.089*

.082**

.078

.0

53*

.069*

(0.063)

(0.067)

(0.048)

(0.180)

(0.0

75)

(0.081)

Firmsinopera

tion(squaredyears)

Firmsize(1=

MSMEs)

-1.231***

-1.028***

-1.345

* * *

-.945**

-.746

-1.0

02***

(0.000)

(0.003)

(0.000)

(0.028)

(0.137)

(0.0

04)

Firmtype(1=

domestic)

-.732**

-.610

-.737**

-.509

-.8

67**

1.085**

(0.019)

(0.149)

(0.032)

(0.275)

(0.0

11)

(0.031)

Location(1=P

hnomPenh)

.157

0.022

.157

-.023

.273

.175

-.067

-.672

-.298

-.827

(0.677)

(0.963)

(0.693)

(0.964)

(0.510)

(0.745)

(0.887)

(0.341)

(0.585)

(0.292)

Sector(1=ma

nufacturing)

.913**

.953*

1.195***

1.098*

(0.023)

(0.084)

(0.002)

(0.098)

Principalalsomanager(1=yes)

.963*

.100

2.101***

.154

1.0

58*

.829

(0.098)

(0.933)

(0.008)

(0.806)

(0.0

62)

(0.410)

Largestshareholderisfamily

-.211

.032

.008

.154

(1=family)

(0.553)

(0.956)

(0.983)

(0.806)

Exportingfirms(export=1)

.678*

.367

(0.081)

(0.482)



177

Table10.


C

ovariates

(1)

(2)

(3)

(4)

(5)

(6)

Pooled

MSME

Pooled

MSME

Pooled

MSME

Pooled

M

SME

Pooled

MSME

Po

oled

MSME

Productassha

reofnational

.030**

.057***

.026*

.048**

.0

26**

.031*

market(%)

(0.025)

(0.007)

(0.103)

(0.050)

(0.0

26)

(0.059)

Totalfemalepermanentworkers(%)

.015**

-.003

.016**

-.005

(0.026)

(0.762)

(0.054)

(0.688)

Productassha

reoflocalmarket(%)

-.035***

-.047**

-.034**

-.045**

.0

31***

-.031

(0.004)

(0.013)

(0.017)

(0.031)

(0.0

03)

(0.039)

Ownorsharegenerator(1=yes)

.634

*

.839

.509

.709

(0.079)

(0.131)

(0.214)

(0.243)

Mobilephoneuseinbusiness

-.826

.265

-.666

1.078

(1=yes)

(0.298)

(0.849)

(0.471)

(0.541)

Workforceuse

computer(%)

.006

.007

.006

.005

(0.242)

(0.369)

(0.307)

(0.546)

E-mailuseinb

usiness(1=yes)

.798

.390

.534

.130

(0.206)

(0.615)

(0.438)

(0.874)

Workforcebelongtounion(%)

.022***

0.037**

.022***

.019

.013*

.031*

.003

.021

0.0

18**

.027*

(0.001)

(0.019)

(0.003)

(0.301)

(0.079)

(0.102)

(0.682)

(0.394)

(0.0

12)

(0.059)

Informationon

firms’competitors,

1.131***

1.643***

1.287***

1.65***

.977**

1.295**

.807

1.316*

.7

88**

.791

uppliersandcustomers

(0.001)

(0.001)

(0.000)

(0.003)

(0.014)

(0.046)

(0.084)

(0.070)

(0.0

16)

(0.114)

#domesticcompetitors

(1=lessthan5)



178

Table10.


Variables

(1)

(2)

(3)

(4)

(5)

(6)

Pooled

MSME

Pooled

MSME

Pooled

MSME

Pooled

M

SME

Pooled

MSME

Pooled

MSME

#foreigncomp

etitors

-.747**

-1.054**

-.987***

-1.20**

-.148

-.296

-.412

-.237

(1=lessthan5)

(0.015)

(0.018)

(0.003)

(0.014)

(0.681)

(0.608)

(0.322)

(0.709)

#domesticsuppliers

-.745**

-1.174***

-.395

-1.10**

-.360

-1.171**

-.345

-1.32**

(1=lessthan5)

(0.019)

(0.006)

(0.227)

(0.016)

(0.348)

(0.039)

(0.433)

(0.033)

#foreignsupp

liers(1=lessthan5)

-.054

-.169

-.068

-.433

(0.898)

(0.788)

(0.889)

(0.517)

#domesticcustomers

-.171

.051

-.413

-.113

(1=lessthan5)

(0.671)

(0.929)

(0.354)

(0.853)

Firms’salessolddomestically(%)

.003

(0.791)

Firms’salesexporteddirectly(%)

.002

-.001

.006

(0.633)

(0.967)

(0.228)

Firms’financia

linformation

Financialstate

mentreviewed

.278

.357

.288

.039

(1=yes)

(0.400)

(0.500)

(0.436)

(0.947)

Firms’termloa

ns(1=yes)

.830***

.955**

1.105***

1.178***

.858***

1.12***

.840**

1.111***

1.171***

1.342**

.7

71**

1.089**

(0.006)

(0.015)

(0.000)

(0.005)

(0.008)

(0.010)

(0.024)

(0.035)

(0.007)

(0.022)

(0.0

23)

(0.022)

Informationan

dperceptions

ongovernmen

tservicesdelivery

andothers

Investmentpro

spects

.282

.611

.555

.602

(1=optimistic)

(0.553)

(0.449)

(0.321)

(0.533)

Legalsystem(1=majorobstacle)

.243

.857

.345

.859

(0.496)

(0.112)

(0.392)

(0.147)



179

Table10.


C

ovariates

(1)

(2)

(3)

(4)

(5)

(6)

Pooled

MSME

Pooled

MSME

Pooled

MSME

Pooled

M

SME

Pooled

MSME

Po

oled

MSME

Viewongov.s

ervices(1=efficient)

-.451

-.805

-.475

-.537

(0.271)

(0.235)

(0.291)

(0.429)

Informationon

SEZ(1=yes)

1.119***

1.501***

.953***

1.518***

.876***

1.384***

.581*

1.285**

.751**

1.458**

.7

80***

1.267***

(0.000)

(0.000)

(0.001)

(0.000)

(0.003)

(0.002)

(0.073)

(0.016)

(0.042)

(0.012)

(0.0

08)

(0.004)

#licenses,per

mits,registrations

.204***

.357***

.261***

.367***

.2

50***

.427***

(0.000)

(0.002)

(0.000)

(0.004)

(0.0

00)

(0.000)

Seniormanagementtime

.002

-.013

-.002

-0.016

dealingwithgo

vernment

(0.790)

(0.268)

(0.823)

(0.187)

requirement(%

)

Firms’disputesoverpayments

-.026

.022

.021

.081

solvedbycour

t(%)

(0.573)

(0.628)

(0.696)

(0.332)

Firmsseektolobby

.077

.276

.182

.559

government(1

=yes)

(0.839)

(0.641)

(0.681)

(0.398)

Inspectionand

meetingwith

.509

.272

.569

.257

gov.officials(1

=yes)

(0.141)

(0.605)

(0.128)

(0.637)

Physicalcapita

l

Hypotheticalvaluesofassets

.261***

.281***

.160*

.078

(machinery,ve

hicles,equipment,

(0.000)

(0.005)

(0.082)

(0.554)

landifowned,

andbuildings

ifowned($)

Ownedland(1

=yes)

-.055

0.110

-.845**

-.627

-1.019**

-.586

(0.861)

(0.796)

(0.016)

(0.196)

(0.030)

(0.390)



180

Table10.


(1)

(2)

(3)

(4)

(5)

(6)

Pooled

MSME

Pooled

MSME

Pooled

MSME

Pooled

M

SME

Pooled

MSME

Pooled

MSME

Education

Firms’formalb

eyond-the-job

.512**

.719**

.614**

.789**

.508*

.959**

.517

1.091

*

training(1=ye

s)

(0.050)

(0.053)

(0.030)

(0.049)

(0.082)

(0.023)

(0.159)

(0.064)

Employeeswit

huniversity/institution

-.009

-.005

levelofeducat

ion(%)

(0.187)

(0.596)

_Cons

-1.573***

-3.159***-1.371***

-2.848***

-4.19***

-6.187***

-4.967***-6.186***-8.119***

-8.56**

-3.2

39***-5.280***

(0.002)

(0.000)

(0.010)

(0.000)

(0.000)

(0.000)

(0.001)

(0.010)

(0.000)

(0.013)

(0.0

00)

(0.000)

LRChi 2

197.870

38.26

218.73

61.34

196.48

61.76

225.20

97.68

224.12

90.43

208.40

81.72

Prob>Chi 2

0.0000

0.0000

0.0000

0.0000

0.0000

0.0000

0.0000

0

.0000

0.0000

0.0000

0.000

0.0000

PseudoR2

0.3340

0.1467

0.3754

0.2395

0.3759

0.2651

0.4347

0

.4141

0.4807

0.4228

0.3

979

0.3385

Obs.

502

350

496

347

448

319

447

315

406

291

453

321

BalancingPropertySatisfied

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

S o u r c e :

Auth

or’scalculation.Figuresinthebracketsarep-values.***indicatesthatthecoefficientsaresign

ificantat1%level,**at5%leveland*at10%

level.



181

Table11.PSM

estimationres

ultsofmembershipsinbu

sinessassociationsorcha

mbersofcommerce(poole

d)

Outco

mevariables

(1)

(2)

(3)

(4)

(5)

(6)

NN

Kernel

NN

Kernel

NN

Kernel

NN

K

ernel

NN

Kernel

N

N

Kernel

Totalsales(lo

g)

0.25

0.33*

0.32

0.39

0.62

0.39***

0.44

-0.03

0.83

-0.02

0

.13

-0.01

(0.69)

(1.72)

(0.74)

(1.54)

(1.29)

(2.47)

(0.85)(-0.08)

(1.42)

(-0.05)

(0

.28)

(-0.02)

Totalproduction(log)

0.34

0.43*

0.39

0.45***

0.70*

0.42***

0.59

0.03

0.79

0.11

0

.20

0.07

(1.01)

(1.86)

(0.99)

(2.23)

(1.67)

(2.09)

(1.26)

(0.10)

(1.47)

(0.34)

(0

.49)

(0.23)

Labourproductivity(log)

0.07

-0.02

0.14

0.08

0.08

0.07

0.11

-0.05

0.04

0.18

-0

.13

-0.04

(0.31)

(-0.09)

(0.58)

(0.37)

(0.31)

(0.43)

(0.42)(-0.17)

(0.14)

(0.71)

(-0

.51)

(-0.18)

Costoflabour(log)

0.43

0.48***

0.24

0.41***

0.60

0.39*

0.50

0.10

0.70

-0.02

0

.33

0.09

(1.45)

(2.88)

(0.67)

(2.22)

(1.56)

(1.67)

(1.15)

(0.50)

(1.44)

(-0.08)

(0

.89)

(0.40)

Totalnumber

ofworkers(log)

0.32

0.52***

0.28

0.43***

0.63*

0.43*

0.50

0.10

0.76*

-0.06

0

.37

0.14

(1.25)

(3.37)

(0.82)

(2.12)

(1.76)

(1.90)

(1.30)

(0.39)

(1.81)

(-0.27)

(1

.14)

(0.61)

Labourcostp

erworker(log)

0.11

-0.01

-0.03

-0.01

-0.03

-0.03

0.01

0.01

-0.06

0.05

-0

.04

-0.04

(0.79)

(-0.09)

(-0.17)

(-0.05)

(0.20)

(-0.23)

(0.04)

(0.03)

(-0.28)

(0.28)

(-0

.27)

(-0.27)

Othercostof

production(log)

1.44

2.66***

1.79

2.28*

3.48*

1.66*

0.50

-0.69

-0.14

-1.75

0

.22

-0.32

(0.92)

(2.43)

(1.01)

(1.79)

(1.78)

(1.71)

(0.23)(-0.46)

(-0.05)

(-0.87)

(0

.12)

(-0.25)

S o u r c e :

Auth

or’scalculation.Figuresinbrack

etsaret-statistics.***indicatesthatcoefficientsaresignificantat1%level,**at5%leveland*at

10%level.



182

T

able12.PSM

estimationre

sultsofmembershipsinbu

sinessassociationsorchambersofcommerce(SME

s)

Outco

mevariables

(1)

(2)

(3)

(4)

(5)

(6)

NN

Kernel

NN

Kernel

NN

Kernel

NN

K

ernel

NN

Kernel

N

N

Kernel

Totalsales(log

)

0.70***

1.00***

0.97***

1.00***

0.76

0.81***

0.78

1.12***

0.54

0.95***

0.20

0.92**

(2.39)

(4.69)

(2.36)

(2.84)

(1.46)

(2.58)

(1.21)

(

2.93)

(0.96)

(2.63)

(0

.37)

(2.13)

Totalproductio

n(log)

0.46*

0.82***

0.97***

0.87***

0.46

0.73***

0.58

0.89***

0.48

0.79*

0.17

0.78**

(1.73)

(3.24)

(2.59)

(2.66)

(1.32)

(2.60)

(1.08)

(

2.22)

(0.90)

(1.98)

(0

.32)

(2.38)

Labourproduc

tivity(log)

0.22

0.20

0.65***

0.34

0.83***

0.61*

0.20

0.32

0.57*

0.37

0.01

0.20

(1.19)

(0.77)

(2.75)

(1.25)

(2.98)

(1.91)

(0.58)

(

0.79)

(1.85)

(0.96)

(0

.02)

(0.55)

Costoflabour

(log)

0.68***

0.82***

0.43

0.64***

0.09

0.35

0.53

0.47*

-0.32

0.36

0.29

0.63**

(2.70)

(3.09)

(1.21)

(2.68)

(0.21)

(1.16)

(0.97)

(

1.70)

(-0.65)

(1.06)

(0

.64)

(2.36)

Totalnumbero

fworkers(log)

0.42***

0.66***

0.34

0.56***

-0.13

0.17

0.42

0.60***

0.001

0.46**

0.21

0.61**

(2.07)

(3.89)

(1.08)

(2.82)

(-0.33)

(0.88)

(0.93)

(

2.66)

(0.01)

(2.11)

(0.51)

(2.18)

Labourcostpe

rworker(log)

0.21

0.17

0.04

0.11

0.16

0.19

0.06

-

0.13

-0.26

-0.10

0.12

0.02

(1.67)

(1.11)

(0.28)

(0.57)

(0.84)

(1.25)

(0.28)

(-

0.71)

(-1.21)

(-0.49)

(0.62)

(0.10)

Othercostofp

roduction(log)

4.46***

5.27***

1.87

4.46***

-0.11

2.54**

2.52

2.35

0.98

2.99*

0.86

2.49

(3.41)

(5.45)

(1.13)

(2.68)

(-0.05)

(2.14)

(0.96)

(

1.23)

(0.38)

(1.91)

(0

.35)

(1.07)

S o u r c e :

Auth


etsaret-statistics.***indicatesthatcoefficientsaresignificantat

1%level,**at5%leveland*at

10%level.



183

Table13.OLSestimationresultsofmembershipsi

nbusinessassociationsorchambersofcommerce

Outco

mevariables

(1)

(2)

(3)

(4)

(5)

(6)

Pooled

SME

Pooled

SME

Pooled

SME

Pooled

SME

Pooled

SME

Po

oled

SME

Totalsales(log

)

0.53***

0.91***

0.49***

0.74***

0.46***

0.45*

0.36*

0.49*

0.23

0.29

0.42**

0.65**

(2.83)

(3.43)

(2.59)

(2.70)

(2.40)

(1.67)

(1.83)

(

1.78)

(1.15)

(1.04)

(2.06)

(2.26)

Totalproductio

n(log)

0.54***

0.73***

0.51***

0.64***

0.49***

0.47*

0.31

0.31

0.22

0.19

0.43**

0.58**

(2.82)

(2.70)

(2.62)

(2.31)

(2.53)

(1.72)

(1.55)

(

1.08)

(1.05)

(0.68)

(2.11)

(1.98)

Labourproduc

tivity(log)

0.13

0.16

0.18

0.24

0.14

0.26

0.06

0.12

0.09

0.11

0.09

0.23

(0.82)

(0.72)

(1.08)

(1.03)

(0.40)

(1.08)

(0.33)

(

0.49)

(0.49)

(0.44)

(0.57)

(0.93)

Costoflabour

(log)

0.53***

0.73***

0.40***

0.48**

0.34**

0.24

0.278*

0.09

0.19

-0.04

0.37**

0.39*

(3.54)

(3.39)

(2.64)

(2.21)

(2.15)

(1.08)

(1.72)

(

0.42)

(1.12)

(-0.18)

(2.30)

(1.65)

Totalnumbero

fworkers(log)

0.46***

0.63***

0.37***

0.45***

0.38***

0.25*

0.30***

0.25*

0.18

0.14

0.38***

0.42***

(4.50)

(4.46)

(3.58)

(3.17)

(3.53)

(1.80)

(2.74)

(

1.68)

(1.58)

(1.00)

(3.29)

(2.66)

Labourcostpe

rworker(log)

0.07

0.12

0.05

0.05

-0.04

0.004

-0.02

-

0.15

0.01

-0.18

0.005

-0.02

(0.77)

(0.84)

(0.43)

(0.36)

(-0.32)

(0.03)

(-0.21)

(-

0.98)

(0.12)

(-1.13)

(0.05)

(-0.14)

Othercostofp

roduction(log)

3.32***

4.86***

3.08***

4.28***

2.29***

2.95***

1.91*

1.66

1.00

1.68

2.31**

2.68*

(3.53)

(3.61)

(3.58)

(3.46)

(2.70)

(2.47)

(1.84)

(

1.11)

(1.14)

(1.36)

(2.23)

(1.81)

S o u r c e :

Auth


etsaret-statistics.***indicatesthatcoefficientsaresignificantat1%level,**at5%leveland*at

10%level.



184

T

able14.PSM

&OLSestimationresultsofmembershipsinbusinessassociation

sorchambersofcommerc

e

Outcomevariables

(1)

(2)

(3)

(4)

(5)

(6)

Pooled

SME

Pooled

SME

Pooled

SME

Pooled

SME

Pooled

SME

Po

oled

SME

Totalsales(log)

0.52***

0.86***

0.48**

0.72***

0.51***

0.43*

0.39**

0.63**

0.30

0.41

0.44**

0.71**

(2.72)

(3.18)

(2.48)

(2.68)

(2.67)

(1.59)

(1.95)

(2.11)

(1.39)

(1.36)

(2.12)

(2.34)

Totalproductio

n(log)

0.51***

0.67**

0.46**

0.60**

0.52

0.45*

0.36*

0.42

0.31

0.39

0.44**

0.65**

(2.64)

(2.45)

(2.42)

(2.26)

(2.62)

(1.63)

(1.83)

(1.42)

(1.40)

(1.29)

(2.16)

(2.15)

Labourproduc

tivity(log)

0.12

0.10

0.17

0.20

0.16

0.23

0.07

0.23

0.12

0.25

0.09

0.21

(0.73)

(0.44)

(0.99)

(0.88)

(0.95)

(0.99)

(0.43)

(0.94)

(0.61)

(0.93)

(0.49)

(0.83)

Costoflabour

(log)

0.56***

0.71***

0.39**

0.46**

0.35**

0.19

0.34**

0.15

0.24

0.09

0.43***

0.52**

(3.70)

(3.30)

(2.54)

(2.10)

(2.16)

(0.85)

(2.18)

(0.57)

(1.33)

(0.03)

(2.71)

(2.17)

Totalnumbero

fworkers(log)

0.45***

0.63***

0.34***

0.45

0.39***

0.24*

0.32***

0.21

0.21*

0.16

0.40***

0.48***

(4.30)

(4.48)

(3.29)

(3.06)

(3.43)

(1.69)

(2.74)

(1.30)

(1.78)

(1.02)

(3.42)

(3.04)

Labourcostpe

rworker(log)

0.13

0.10

0.06

0.03

-0.02

-0.33

0.02

-0.07

0.02

-0.16

0.05

0.04

(1.20)

(0.71)

(0.56)

(0.25)

(-0.23)

(-0.22)

(0.23)

(-0.43)

(0.16)

(-0.82)

(0.39)

(0.21)

Othercostofp

roduction(log)

3.25***

4.94***

2.87***

4.24***

2.19**

2.84**

2.04*

2.03

0.87

1.53

2.19**

2.64*

(3.49)

(3.59)

(3.32)

(3.50)

(2.48)

(2.27)

(1.89)

(1.24)

(0.91)

(1.15)

(2.10)

(1.68)

Regionofcom

monsupport

[0.0329-

[0.0230-

[0.0222-

[0.0131-

[0.0244-

[0.0116-

[0.0609-[0.0251-

[0.0273-

[0.0206-

[0.0425-

[0.0193-

0.9333]

0.5096]

0.9878]

0.9885]

0.9774]

0.9326]

1.0000]

1.000]

1.000]

1.000]

[1.000]

1.000]

S o u r c e :

Auth

or’scalculation.Figuresinbracketsaret-statistics.***indicatesthatcoefficientsaresignificantat1%level,**at5%leveland*at10%level.



185

Appendix A

Variable definitions

Variable Definition

Sales Firms’ annual sales in USD for fiscal years

2006 and 2005.

Production Firms’ annual production in USD for fiscal

years 2006 and 2005

Cost of labour Firms’ annual cost of labour in USD for

fiscal years 2006 and 2005. The costs

include wages, salaries, bonuses, social

payments, and others.

Labour cost per worker Firm’s cost per workers in USD for fiscal

years 2006 and 2005. This variable is the

division between total labour cost in each

fiscal year and the total permanent workers

firms had in respective fiscal year.

Permanent workers include management,

professionals, skilled production workers,

unskilled production workers, and non-

production/service workers.

Cost of intermediate goods Firms’ annual cost of raw materials and

intermediate goods used in production and

goods and materials purchased for re-sale

in USD for fiscal years 2006 and 2005.

Other costs Firms’ other costs of production in USD for

fiscal years 2006 and 2005 include costs

of electricity, fuel, water, communication

services, transportation, rents, and repairs

and maintenance.Labour productivity Production per worker in USD for fiscal

years 2006 and 2005.

# of regular employees Total number of permanent workers

in fiscal years 2006 and 2005.

# of customers This is defined as total number of

customers within the main product line in

domestic market. Customers include

domestic private firms, Cambodian

state-owned firms, foreign owned firms,

NGOs, and others.



186

Business Association membership A 0/1 dummy variable with “1” representing

firms’ membership in business association

or chamber of commerce and “0” otherwise.

Firm characteristics

Firm age The number of years that firms had been in

operation from establishment to April 2006.

Firms which started operation after April

2006 were terminated.

Investment prospect of firms This was taken from a question “what is

your [firms] prospect for investment in the

next 3 years?” The variable takes the value

of “1” if firms are optimistic and

“0” otherwise”.

Perceptions on legal system A categorical variable having three

or conflict resolution groups: 1 “no obstacle” 2 “minor obstacle”

and 3 “more obstacle”.

Efficiency of government service delivery A categorical variable having three

groups: 1 “Inefficient” 2 “Somewhat

efficient” and 3 “Efficient”.

Senior management time in dealing This is defined, in a typical week, as % of

with government requirements senior management’s time spent in dealing

with government requirements (e.g. taxes,customs, labour regulations, licensing and

registration) which includes dealing with

officials and completing forms.

Government lobby A 0/1 dummy variable that takes value “1” if

firms seek to lobby government or influence

content of laws or regulations affecting firms

and “0” otherwise.

Term loans with financial institution A 0/1 dummy variable taking value “1” if firms has term loan (more than 6 months)

from a bank or financial institution and “0”

otherwise.

Term loans with collateral A 0/1 dummy variable with “1” representing

that the financing requires collateral or a

deposit and 0 otherwise.

Formal training to permanent employees A 0/1 dummy variable having value “1” if

firms offer formal beyond the job training to

permanent employees and “0” otherwise.



187

Permanent skilled employees received This is defined as the percentage of

training permanent skilled staff who received

formal training in 2006.

Permanent unskilled employees This is defined as the percentage of

received training permanent unskilled staff who receivedformal training in 2006.

Number of foreign nationals to Number of skilled workers who were

total employees foreign nationals in 2006.

Information on Special Economic Zones A 0/1 dummy variable taking value “1”

if firms have heard of and “0” otherwise.

Educational levels of employees The overall percentage of the workforce in

the firms who have the following

educational levels: Primary school (belowgrade 6), Up to Lower Secondary (grade

7-9), Up to Upper Secondary (grade

10-12), Up to University and Institutions.

Sector This variable contains four sectors:

Manufacturing (food, textiles, garments,

chemicals, plastics & rubber, basic metals

and fabricated metal products, machinery

and equipment, electronics, and others),

Trade (wholesale including export services,retail), Tourism (hotels and restaurants,

travel agencies, tour operators, etc.), and

Other (construction, transport, IT, etc.).

Firm size This variable defines two groups of firms:

Small and Medium Enterprises and Large

Enterprises. The number of regular

employees firms have in April 2006 and the

definition of firms given by the government

are used to derive these categories. Thisvariable assumes binary value with “1”

representing MSMEs and “0” LEs.

Location Battambang, Siem Reap, Phnom Penh,

and Kampong Cham

Exporting indicators Exporters and non-exporters

Equity stakes of firms Domestically-owned (100%), foreign-owned

(100%), and joint-venture

Source: Author’s preparation



188

Referenc

Aerts, K., and D. Czarnitzki (2004). Using Innovation Survey Data to Evaluate R&D Policy:

The Case of Belgium, Discussion Paper No. 05-55. Centre for European Economic

Research. Mannheim. Aterido, R., M. Hallward-Driemeier and C. Pages (2009). Big Constraints to Small Firms’

Growth? Business Environment and Employment Growth across Firms, World Bank

Policy Research Working Paper 5032. World Bank. Washington D.C.

Ayyagari, M., A. Demirguc-Kunt and V. Maksimovic (2011). Small vs. Young Firms across

the World , World Bank Policy Research Working Paper 5631, World Bank,

Washington, D.C.

Ayyagari, M., A. Demirguc-Kunt and V. Maksimovic (2008). “How Important are Financing

Constraints? The Role of Finance in the Business Environment”, World Bank Economic Review , 22(3), pp. 483-516.

Baily, P., (2007). Cambodian Small and Medium Sized Enterprises: Constraints, Policies

and Proposals for their Development, ERIA Research Report, http://www.eria.org/

research/images/pdf/PDF%20No.5/No,5-1-Cambodian.pdf (accessed October

2011).

Batra, G., and S. Mahmood (2003). Direct Support to Private Firms: Evidence on

Effectiveness, World Bank Policy Research Working Paper 3170, World Bank,

Washington, D.C.

Beck, T., A. Demirguc-Kunt and M.S.M. Peria (2008), Bank Financing for SMEs around the

World: Drivers, Obstacles, Business Models, and Lending Practices, World Bank

Policy Research Working Paper 4785, World Bank, Washington, D.C.

Becker, G.S., (1985). “Public Policies, Pressure Groups, and Dead Weight Costs”, Journal

of Public Economics, 28; pp. 329-347.

Becker, S.O., and A. Ichino (2002). “Estimation of Average Treatment Effects Based on

Propensity Score”, Stata Journal , 2(4); pp. 358-377.

Bennett, R.J., (2008). SME Policy Support in Britain Since the 1990s: What have welearnt? Environment and Planning C: Government and Policy , 26; pp. 375-397.

Bennett, R.J., (2000). “The Logic of Membership of Sectoral Business Associations”,

Review of Social Economy , LVIII(1); pp. 18-42.

Bennett, R.J., (1999); “Business Routes of Influence in Brussels: Exploring the Choice of

Direct Representation”, Political Studies, XLVII; pp. 240-257.

Bennett, R.J., (1998). Business Associations and their Potential Contribution to the

Competitiveness of SMEs, Entrepreneurship & Regional Development , 10;

pp. 243-260.

Bennett, R.J., (1995). The Logic of Local Business Associations: An Analysis of Voluntary

Chambers of Commerce, Journal of Public Policy , 15(3); pp. 251-279.



189

Bennett, R.J., and M. Ramsden (2007). The Contribution of Business Associations to

SMEs: Strategy, Bundling or Reassurance? International Small Business Journal ,

25(1); pp. 49-76.

Bernard, A.B., and J.B. Jensen (1997). “Exporters, Skill Upgrading, and the Wage Gap”,

Journal of International Economics, 42; pp. 3-31.

Biesebroeck, J.V., (2003). Exporting Raises Productivity in Sub-Saharan African

Manufacturing Plants, NBER Working Paper Series, No. 10020, BBER, Cambridge.

Black, D.A., and J.A. Smith (2004). How robust is the evidence on the effects of college

quality? Evidence from matching, Journal of Econometrics, 121; pp. 99-124.

Caliendo, M., and S. Kopeinig (2008). Some Practical Guidance for the Implementation of

Propensity Score Matching, Journal of Economic Surveys, 22(1); pp. 31-72.

Cambodian Federation of Employers and Business Associations (CAMFEBA) (2011). Types

of Membership, http://www.camfeba.com/en/membership/type-of-membership.html

(accessed June 2012).

Committee for Economic and Financial Policy (CEFP) (2011). Review Report on the

Implementation of the Policy Paper on the Promotion of Paddy Production and Rice

Export 2010-2011 (Phnom Penh: CEFP).

Chheang, Vannarith, Oum Sothea and Thearith Leng (2011). Constraints on SMEs in

Cambodia and their Participation in Production Network , http://eria.org/pdf/research/

y2009/no8/CH05_SMEs.pdf (accessed October 2011).

Council for the Development of Cambodia (CDC). (2005). Draft Outline of SME

Development Framework (Phnom Penh: CDC) http://www.mef.gov.kh/documents/

PFM/7cg_document/draft_outline_sme.htm (accessed 2 May 2011).

Criscuolo, C., R. Martin, H. Overman and J.V. Reenen (2007). “The Effects of Industrial

Policy on Corporate Performance: Evidence from Panel Data”, Centre for Economic

Performance, London School of Economics.

Czarnitzki, D., and H. Hottenrott (2011). R&D investment and financing constraint of small

and medium-sized firms, Small Business Economics, 36; pp. 65-83.

Dinh, H.T., A. Mavrids, A. Dimitris and H.B. Nguyen (2010). The Binding Constraint on

Firms’ Growth in Developing Countries, Policy Research Working Paper, No. 5485,

World Bank, Washington, D.C.

Doner, R.F., and B.R. Schneider (2000). Business Associations and Economic

Development: Why Some Associations Contribute More Than Others, Business and

Politics, 2(3); pp.261-288.

Doner, R.F., and B.R. Schneider (1999). Business Associations and Economic

Development , Institute for Policy Research Working Paper WP-99-12, Institute for

Policy Research, North-Western University.



190

The Federation of Association of Small and Medium Enterprises of Cambodia (FASMEC)

(2010). Constitution of the Federation of Association of Small and Medium

Enterprises of Cambodia (Phnom Penh: FASMEC), http://fasmec.com/images/

pictures/FASMEC_Constitution_EN.pdf (accessed 23 November 2011).

Germany Trade & Invest (2012). Investment Guide to Germany , http://www.gtai.de/GTAI/Content/EN/Invest/_SharedDocs/Downloads/GTAI/Brochures/Germany/investment-

guide-to-germany-2012.pdf (accessed July 2012).

Gertler, P.J., S. Martinez, P. Premand, L.B. Rawlings and C.M.J. Vermeersch (2011b).

“Chapter 7: Matching,” in Interactive Textbook on Impact Evaluation in Practice, the

International Bank for Reconstruction and Development; pp.107-116, World Bank,

Washington, D.C.

Goldsmith, A.A., (2000). Business Associations and Better Governance in Africa, Public

Administration and Development , 22; pp. 39-49.Han, L., and A. Benson (2010). The use and usefulness of financial assistance to UK

SMEs, Environment and Planning C: Government and Policy , 28; pp. 552-566.

Harner, S., (2003). Financing SMEs in Cambodia: Why do banks find it so difficult?http://

www.ifc.org/ifcext/mekongpsdf.nsf/AttachmentsByTitle/PSDP+14/$FILE/PSDP-No-

14-EN.pdf (accessed October 2011).

Harvie, C., D. Narjoko and S. Oum (2010). Firm Characteristic Determinants of SME

Participation in Production Network , Discussion Paper Series 11, ERIA, Jakarta.

Heckman, J., (1979). Sample Selection bias as a specification error, Econometrica, 47;

pp. 153-161.

Heckman, J., I. Hidehiko, J. Smith and P. Todd (1998). Characterizing Selection Bias Using

Experimental Data, Econometrica, 66(5); pp. 1017-1098.

Isgut, A.E., (2001). What’s Different about Exporters? Evidence from Colombian

Manufacturing, Journal of Development Studies, 37(5); pp. 57-82.

Khandker, S.R., G.B. Koolwal and H.A. Samad (2010). Chapter 4: Propensity Score

Matching, in Handbook on Impact Evaluation: Quantitative Methods and Practices,

the International Bank for Reconstruction and Development; pp. 53-69, World Bank,

Washington, D.C.

Lee, W.S., (2006). Evaluating the Effects of a Mandatory Government Program using

Matched Groups within a Similar Geographic Location, Melbourne Institute of

Applied Economic and Social Research, the University of Melbourne, Melbourne.

Levitsky, J., (1992). Private sector membership associations and support for SMEs, Small

Enterprises Development , 3(1); pp. 22-34.

Luna, M., and R. Tirado (2008). Business Associations and their contribution to knowledge

networks in Mexico, International Journal of Technology Management and

Sustainable Development , 7(3); pp.251-264.



191

McCormick, L.E., J.D. Hawley and E. Meléndez (2008). The Economic and Workforce

Development Activities of American Business Associations, Economic Development

Quarterly , 22(3); pp. 213-227.

Meas, W.H., (2006). Characteristics of Small and Medium Enterprises in Cambodia: Case

Study of Rice Milling Enterprises, http://www.ncsb2006.se/pdf/Characteristics%20in%20Cambodia.pdf (accessed October 2011).

Mitchell, N.J., (1990). The Decentralization of Business in Britain, Journal of Politics, 52(2);

pp. 622-637.

Mole, K., M. Hart, S. Roper and D. Saal (2008). Differential Gains from Business Link

Support and Advise: A Treatment Effects Approach, EPC: Government and Policy ,

26; pp. 315-334, Pion Publishing, Great Britain.

Moore, M., and L. Hamalai (1993). Economic Liberalization, Political Pluralism, Business

Associations in Developing Countries, World Development , 21(12); pp.1895-1912.

Motohashi, K., (2002). Use of Plant-Level Micro-Data for the Evaluation of SME Innovation

Policy in Japan, Science, Technology and Industry Working Paper 2002/12, OECD,

Paris.

Nadvi, K., (1999). Industrial Clusters and Networks: Case Studies of SME Growth and

Innovation, Small and Medium Enterprise, UNIDO, Vienna.

National Institute of Statistics (NIS) (2011). Preliminary Results of 2011 Economic Census

of Cambodia, MoP, Phnom Penh.

National Institute of Statistics (NIS) (2008). NIS Statistical Yearbook 2008 (Phnom Penh:

MoP).

Oh, Inha, Lee Jeong Dong and Heshmati Almas (2009), Evaluation of credit guarantee

policy using propensity score matching Small Business Economics, 33;

pp. 335-351.

Olson, M., (1965). The Logic of Collective Action, Harvard University Press, Cambridge.

Perry, M., (2007). Trade Associations and Business Associability: Strengthening

contributions to business and economic development, New Zealand Centre for Small and Medium Enterprise Research, Massey University, New Zealand.

Royal Government of Cambodia (RGC) (2005). Small and Medium Enterprise Development

Framework , RGC, Phnom Penh.

Royal Government of Cambodia (RGC) (2004). Rectangular Strategy for Growth,

Employment, Equity and Efficiency in Cambodia, http://www.ilo.org/wcmsp5/groups/

public/—asia/—ro-bangkok/documents/genericdocument/wcms_100515.pdf

(accessed 2 May 2011).

Roper, S., and N. Hewitt-Dundas (2001). Grant Assistance and Small Firm Development inNorthern Ireland and the Republic of Ireland, Scottish Journal of Political Economy ,

48(1); pp. 99-117.



192

Rosenbaum, P.R., and D.B. Rubin (1983). The Central Role of the Propensity Score in

Observational Studies for Causal Effects, Biometrika, 70(1); pp. 41-55.

SME Forum (2010). Minutes of the forum, General Department of Industry, Ministry of

Industry, Mines and Energy, Phnom Penh.

Salisbury, H., (1984). Interest Representation: The Dominance of Institutions, American

Political Science Review , 78(1); pp. 64-76.

Shariff, M.N.M., and C. Peou (2008). The Relationship of Entrepreneurship Values, Firm

Financing and the Management and Growth Performance of Small Medium

Enterprises in Cambodia, Problems and Perspectives in Management , 6(4);

pp. 55-64.

Spence, M.M., (2003). Evaluating Export Promotion Programmes: U.K. Overseas Trade

Missions and Export Performance, Small Business Economics, 20(1); pp. 83-103.

Stephanou, C., and C. Rodriguez (2008). Bank Financing to Small and Medium-Sized

Enterprises (SMEs) in Colombia, World Bank Policy Research Working Paper 4481,

World Bank, Washington, D.C.

Stiglitz, J.E., and A. Weiss (1981). Credit Rationing in Markets with Imperfect Information”,

American Economic Review , 71(3); pp. 393-410.

Sukiassyan, G., and J.B. Nugent. (2008). Associations versus Registration as Alternative

Strategies of Small Firms, Small Business Economics, 31; pp. 147-161.

Wilts, A., and M. Meyer (2005). Small Firm Membership in National Associations, Journal of

Public Affairs, 5(2); pp. 176-185.

World Bank (2012). Doing Business in a More Transparent World , World Bank,

Washington, D.C. http://www.doingbusiness.org/~/media/FPDKM/Doing%

20Business/Documents/Annual-Reports/English/DB12-FullReport.pdf (accessed

July 2012).

World Bank (2010). Evaluating SME Support Programs in Chile, in Gladys Lopez-Acevedo

& Hong W. Tan. (eds.), Impact Evaluation of Small and Medium Enterprise

Programs in Latin America and the Caribbean; pp.33-55, World Bank, Washington,

D.C.

World Bank (2007). World Bank Enterprise Surveys: Cambodia. https://www.enterprise

surveys.org/Portal/elibrary.aspx?libid=14 (accessed 6 June 2011).

World Bank (2007).Sampling Procedure Cambodia, http://microdata.worldbank.org/

index.php/catalog/634 (accessed August 2011).

Wren, C., and D.J. Storey (2002). Evaluating the Effect of Soft Business Support upon

Small and Firm Performance, Oxford Economic Papers, 54; pp. 334-365.

Zecchini, S., and M. Ventura (2009). The Impact of Public Guarantees on Credit to SMEs,

Small Business Economics, 32; pp. 191-206.

Zhao, Z., (2005). Sensitivity of Propensity Score Methods to the Specifications, Discussion

Paper Series No. 1873, IZA, Bonn.



193

ARTNeT publishes a wide range of publications dealing withvarious aspects of trade and investment

Policy Brief Series

Duty-Free, Quota-Free Trade for Asia-Pacific Least Developed Countries: Overview and

Update

Pedro J. Martinez Edo and Adam Heal, Brief No. 36, September 2013

Remittances, Migration and Inclusive Growth: The Case of Nepal

Nephil Matangi Maskay and Shiva Raj Adhikari, Brief No. 35, September 2013

Working Paper Series

Addressing Non-tariff Measures in ASEANGloria O. Pasadilla, ARTNeT Working Paper Series, No. 130, September 2013

Who Profits from Trade Facilitation Initiatives?

Bernard Hoekman and Ben Shepherd, ARTNeT Working Paper Series, No. 129,

September 2013

Books, research reports, briefing papers

A Handbook on Negotiating Preferential Trade Agreements: Services Liberalization

Pierre Sauvé and Simon Lacey

ARTNeT Census Briefing Paper on the European Union’s New Timber Regulations and the

Asia-Pacific

Teemu Alexander Puutio, Issue 1, March 2013

The Gravity Model of International Trade: A User Guide

Ben Shepherd, (ST/ESCAP/2645). 2012

For more information visit us at www.artnetontrade.org